Mercurial > repos > deepakjadmin > mayatool3_test3
comparison mayachemtools/docs/scripts/html/code/InfoTextFiles.html @ 0:73ae111cf86f draft
Uploaded
author | deepakjadmin |
---|---|
date | Wed, 20 Jan 2016 11:55:01 -0500 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:73ae111cf86f |
---|---|
1 <html> | |
2 <head> | |
3 <title>MayaChemTools:Code:InfoTextFiles.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: InfoTextFiles.pl,v $</span> | |
17 4 <span class="c"># $Date: 2015/02/28 20:46:20 $</span> | |
18 5 <span class="c"># $Revision: 1.30 $</span> | |
19 6 <span class="c">#</span> | |
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span> | |
21 8 <span class="c">#</span> | |
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span> | |
23 10 <span class="c">#</span> | |
24 11 <span class="c"># This file is part of MayaChemTools.</span> | |
25 12 <span class="c">#</span> | |
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span> | |
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span> | |
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span> | |
29 16 <span class="c"># later version.</span> | |
30 17 <span class="c">#</span> | |
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span> | |
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span> | |
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span> | |
34 21 <span class="c"># details.</span> | |
35 22 <span class="c">#</span> | |
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span> | |
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span> | |
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span> | |
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span> | |
40 27 <span class="c">#</span> | |
41 28 | |
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span> | |
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$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">FileUtil</span><span class="sc">;</span> | |
49 36 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span> | |
50 37 | |
51 38 <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> | |
52 39 | |
53 40 <span class="c"># Autoflush STDOUT</span> | |
54 41 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span> | |
55 42 | |
56 43 <span class="c"># Starting message...</span> | |
57 44 <span class="i">$ScriptName</span> = <span class="i">basename</span><span class="s">(</span><span class="i">$0</span><span class="s">)</span><span class="sc">;</span> | |
58 45 <span class="k">print</span> <span class="q">"\n$ScriptName: Starting...\n\n"</span><span class="sc">;</span> | |
59 46 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span> | |
60 47 | |
61 48 <span class="c"># Get the options and setup script...</span> | |
62 49 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
63 50 <span class="k">if</span> <span class="s">(</span><span class="i">$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> | |
64 51 <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> | |
65 52 <span class="s">}</span> | |
66 53 | |
67 54 <span class="k">my</span><span class="s">(</span><span class="i">@TextFilesList</span><span class="s">)</span><span class="sc">;</span> | |
68 55 <span class="i">@TextFilesList</span> = <span class="i">ExpandFileNames</span><span class="s">(</span>\<span class="i">@ARGV</span><span class="cm">,</span> <span class="q">"csv tsv"</span><span class="s">)</span><span class="sc">;</span> | |
69 56 | |
70 57 <span class="c"># Process options...</span> | |
71 58 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span> | |
72 59 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span> | |
73 60 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
74 61 | |
75 62 <span class="k">print</span> <span class="q">"Checking input text file(s)...\n"</span><span class="sc">;</span> | |
76 63 <span class="k">my</span><span class="s">(</span><span class="i">%TextFilesInfo</span><span class="s">)</span><span class="sc">;</span> | |
77 64 <span class="i">RetrieveTextFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
78 65 <span class="i">ProcessColumnsInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
79 66 | |
80 67 <span class="c"># Generate output files...</span> | |
81 68 <span class="k">my</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | |
82 69 <span class="k">if</span> <span class="s">(</span><span class="i">@TextFilesList</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | |
83 70 <span class="k">print</span> <span class="q">"\nProcessing text files...\n"</span><span class="sc">;</span> | |
84 71 <span class="s">}</span> | |
85 72 <span class="k">for</span> <span class="i">$FileIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span> | |
86 73 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$FileIndex</span>]<span class="s">)</span> <span class="s">{</span> | |
87 74 <span class="k">print</span> <span class="q">"\nProcessing file $TextFilesList[$FileIndex]...\n"</span><span class="sc">;</span> | |
88 75 <span class="i">ListTextFileInfo</span><span class="s">(</span><span class="i">$FileIndex</span><span class="s">)</span><span class="sc">;</span> | |
89 76 <span class="s">}</span> | |
90 77 <span class="s">}</span> | |
91 78 <span class="i">ListTotalSizeOfFiles</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
92 79 | |
93 80 <span class="k">print</span> <span class="q">"\n$ScriptName:Done...\n\n"</span><span class="sc">;</span> | |
94 81 | |
95 82 <span class="i">$EndTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span> | |
96 83 <span class="i">$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> | |
97 84 <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">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span> | |
98 85 | |
99 86 <span class="c">###############################################################################</span> | |
100 87 | |
101 88 <span class="c"># List appropriate information...</span> | |
102 <a name="ListTextFileInfo-"></a> 89 <span class="k">sub </span><span class="m">ListTextFileInfo</span> <span class="s">{</span> | |
103 90 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | |
104 91 <span class="k">my</span><span class="s">(</span><span class="i">$TextFile</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$LineCount</span><span class="cm">,</span> <span class="i">$EmptyLinesCount</span><span class="cm">,</span> <span class="i">$EmptyColDataLinesCount</span><span class="cm">,</span> <span class="i">$GreaterThanMaxColLinesCount</span><span class="cm">,</span> <span class="i">$Label</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$EmptyColValueFound</span><span class="cm">,</span> <span class="i">$PrintTextLine</span><span class="cm">,</span> <span class="i">$NonNumericalDataFound</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">%EmptyColValuesCountMap</span><span class="cm">,</span> <span class="i">%NonEmptyColValuesCountMap</span><span class="cm">,</span> <span class="i">%SpecifiedNonNumericalColValuesCountMap</span><span class="cm">,</span> <span class="i">%NonNumericalColValuesCountMap</span><span class="cm">,</span> <span class="i">%NumericalColValuesCountMap</span><span class="cm">,</span><span class="s">)</span><span class="sc">;</span> | |
105 92 | |
106 93 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span> | |
107 94 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>]<span class="sc">;</span> | |
108 95 <span class="i">@ColLabels</span> = <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]}<span class="sc">;</span> | |
109 96 | |
110 97 <span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">"$TextFile"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Can't open $TextFile: $! \n"</span><span class="sc">;</span> | |
111 98 | |
112 99 <span class="i">$LineCount</span> = <span class="n">0</span><span class="sc">;</span> | |
113 100 <span class="i">$EmptyLinesCount</span> = <span class="n">0</span><span class="sc">;</span> | |
114 101 <span class="i">$EmptyColDataLinesCount</span> = <span class="n">0</span><span class="sc">;</span> | |
115 102 <span class="i">$GreaterThanMaxColLinesCount</span> = <span class="n">0</span><span class="sc">;</span> | |
116 103 | |
117 104 <span class="i">%EmptyColValuesCountMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
118 105 <span class="i">%NonEmptyColValuesCountMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
119 106 <span class="i">%SpecifiedNonNumericalColValuesCountMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
120 107 <span class="i">%NonNumericalColValuesCountMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
121 108 <span class="i">%NumericalColValuesCountMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
122 109 | |
123 110 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">ParseLines</span>}<span class="s">)</span> <span class="s">{</span> | |
124 111 <span class="c"># Skip over column labels from old file...</span> | |
125 112 <span class="k">if</span> <span class="s">(</span><span class="q"><TEXTFILE></span><span class="s">)</span> <span class="s">{</span> | |
126 113 <span class="i">$LineCount</span>++<span class="sc">;</span> | |
127 114 <span class="j">LINE:</span> <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="q"><TEXTFILE></span><span class="s">)</span> <span class="s">{</span> | |
128 115 <span class="i">$LineCount</span>++<span class="sc">;</span> | |
129 116 <span class="i">$PrintTextLine</span> = <span class="n">0</span><span class="sc">;</span> | |
130 117 <span class="i">$Line</span> =~ <span class="q">s/(\r\n)|(\r)|\n//g</span><span class="sc">;</span> | |
131 118 <span class="i">@LineWords</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span> | |
132 119 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CountEmpty</span>}<span class="s">)</span> <span class="s">{</span> | |
133 120 <span class="c"># Count lines with no data...</span> | |
134 121 <span class="k">if</span> <span class="s">(</span>!<span class="i">@LineWords</span><span class="s">)</span> <span class="s">{</span> | |
135 122 <span class="i">$EmptyLinesCount</span>++<span class="sc">;</span> | |
136 123 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">2</span><span class="s">)</span> <span class="s">{</span> | |
137 124 <span class="k">print</span> <span class="q">"Line number $LineCount is empty...\n"</span><span class="sc">;</span> | |
138 125 <span class="s">}</span> | |
139 126 <span class="k">next</span> <span class="j">LINE</span><span class="sc">;</span> | |
140 127 <span class="s">}</span> | |
141 128 <span class="c"># Count lines with empty data for some columns...</span> | |
142 129 <span class="i">$EmptyColValueFound</span> = <span class="n">0</span><span class="sc">;</span> | |
143 130 <span class="j">VALUE:</span> <span class="k">for</span> <span class="i">$Value</span> <span class="s">(</span><span class="i">@LineWords</span><span class="s">)</span> <span class="s">{</span> | |
144 131 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$Value</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
145 132 <span class="i">$EmptyColValueFound</span> = <span class="n">1</span><span class="sc">;</span> | |
146 133 <span class="k">next</span> <span class="j">VALUE</span><span class="sc">;</span> | |
147 134 <span class="s">}</span> | |
148 135 <span class="s">}</span> | |
149 136 <span class="k">if</span> <span class="s">(</span><span class="i">$EmptyColValueFound</span><span class="s">)</span> <span class="s">{</span> | |
150 137 <span class="i">$EmptyColDataLinesCount</span>++<span class="sc">;</span> | |
151 138 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">2</span><span class="s">)</span> <span class="s">{</span> | |
152 139 <span class="k">print</span> <span class="q">"Line number $LineCount contains empty column value(s)...\n"</span><span class="sc">;</span> | |
153 140 <span class="s">}</span> | |
154 141 <span class="i">$PrintTextLine</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">3</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
155 142 <span class="s">}</span> | |
156 143 <span class="c"># Count lines with columns greater than the column label line...</span> | |
157 144 <span class="k">if</span> <span class="s">(</span><span class="i">@LineWords</span> > <span class="i">@ColLabels</span><span class="s">)</span> <span class="s">{</span> | |
158 145 <span class="i">$GreaterThanMaxColLinesCount</span>++<span class="sc">;</span> | |
159 146 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">2</span><span class="s">)</span> <span class="s">{</span> | |
160 147 <span class="k">print</span> <span class="q">"Line number $LineCount contains more than "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@ColLabels</span><span class="s">)</span><span class="cm">,</span> <span class="q">" columns...\n"</span><span class="sc">;</span> | |
161 148 <span class="s">}</span> | |
162 149 <span class="i">$PrintTextLine</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">3</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
163 150 <span class="s">}</span> | |
164 151 <span class="c"># Count empty values for each coulmn...</span> | |
165 152 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#LineWords</span><span class="s">)</span> <span class="s">{</span> | |
166 153 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@ColLabels</span><span class="s">)</span> <span class="s">{</span> | |
167 154 <span class="i">$Label</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span> | |
168 155 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$LineWords</span>[<span class="i">$ColNum</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
169 156 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$NonEmptyColValuesCountMap</span>{<span class="i">$Label</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
170 157 <span class="i">$NonEmptyColValuesCountMap</span>{<span class="i">$Label</span>} += <span class="n">1</span><span class="sc">;</span> | |
171 158 <span class="s">}</span> | |
172 159 <span class="k">else</span> <span class="s">{</span> | |
173 160 <span class="i">$NonEmptyColValuesCountMap</span>{<span class="i">$Label</span>} = <span class="n">1</span><span class="sc">;</span> | |
174 161 <span class="s">}</span> | |
175 162 <span class="s">}</span> | |
176 163 <span class="k">else</span> <span class="s">{</span> | |
177 164 <span class="i">$PrintTextLine</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">3</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
178 165 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$EmptyColValuesCountMap</span>{<span class="i">$Label</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
179 166 <span class="i">$EmptyColValuesCountMap</span>{<span class="i">$Label</span>} += <span class="n">1</span><span class="sc">;</span> | |
180 167 <span class="s">}</span> | |
181 168 <span class="k">else</span> <span class="s">{</span> | |
182 169 <span class="i">$EmptyColValuesCountMap</span>{<span class="i">$Label</span>} = <span class="n">1</span><span class="sc">;</span> | |
183 170 <span class="s">}</span> | |
184 171 <span class="s">}</span> | |
185 172 <span class="s">}</span> | |
186 173 <span class="s">}</span> | |
187 174 <span class="s">}</span> | |
188 175 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CheckData</span>}<span class="s">)</span> <span class="s">{</span> | |
189 176 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#LineWords</span><span class="s">)</span> <span class="s">{</span> | |
190 177 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@ColLabels</span><span class="s">)</span> <span class="s">{</span> | |
191 178 <span class="k">if</span> <span class="s">(</span><span class="i">IsNumerical</span><span class="s">(</span><span class="i">$LineWords</span>[<span class="i">$ColNum</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
192 179 <span class="i">$Label</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span> | |
193 180 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$NumericalColValuesCountMap</span>{<span class="i">$Label</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
194 181 <span class="i">$NumericalColValuesCountMap</span>{<span class="i">$Label</span>} += <span class="n">1</span><span class="sc">;</span> | |
195 182 <span class="s">}</span> | |
196 183 <span class="k">else</span> <span class="s">{</span> | |
197 184 <span class="i">$NumericalColValuesCountMap</span>{<span class="i">$Label</span>} = <span class="n">1</span><span class="sc">;</span> | |
198 185 <span class="s">}</span> | |
199 186 <span class="s">}</span> | |
200 187 <span class="k">else</span> <span class="s">{</span> | |
201 188 <span class="i">$Label</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span> | |
202 189 <span class="k">if</span> <span class="s">(</span><span class="i">IsNotEmpty</span><span class="s">(</span><span class="i">$LineWords</span>[<span class="i">$ColNum</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
203 190 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$NonNumericalColValuesCountMap</span>{<span class="i">$Label</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
204 191 <span class="i">$NonNumericalColValuesCountMap</span>{<span class="i">$Label</span>} += <span class="n">1</span><span class="sc">;</span> | |
205 192 <span class="s">}</span> | |
206 193 <span class="k">else</span> <span class="s">{</span> | |
207 194 <span class="i">$NonNumericalColValuesCountMap</span>{<span class="i">$Label</span>} = <span class="n">1</span><span class="sc">;</span> | |
208 195 <span class="s">}</span> | |
209 196 <span class="s">}</span> | |
210 197 <span class="s">}</span> | |
211 198 <span class="s">}</span> | |
212 199 <span class="s">}</span> | |
213 200 <span class="s">}</span> | |
214 201 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CheckNumericalData</span>}<span class="s">)</span> <span class="s">{</span> | |
215 202 <span class="i">$NonNumericalDataFound</span> = <span class="n">0</span><span class="sc">;</span> | |
216 203 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColNums</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span> | |
217 204 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@LineWords</span><span class="s">)</span> <span class="s">{</span> | |
218 205 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsNumerical</span><span class="s">(</span><span class="i">$LineWords</span>[<span class="i">$ColNum</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
219 206 <span class="i">$NonNumericalDataFound</span> = <span class="n">1</span><span class="sc">;</span> | |
220 207 <span class="i">$Label</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span> | |
221 208 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$SpecifiedNonNumericalColValuesCountMap</span>{<span class="i">$Label</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
222 209 <span class="i">$SpecifiedNonNumericalColValuesCountMap</span>{<span class="i">$Label</span>} += <span class="n">1</span><span class="sc">;</span> | |
223 210 <span class="s">}</span> | |
224 211 <span class="k">else</span> <span class="s">{</span> | |
225 212 <span class="i">$SpecifiedNonNumericalColValuesCountMap</span>{<span class="i">$Label</span>} = <span class="n">1</span><span class="sc">;</span> | |
226 213 <span class="s">}</span> | |
227 214 <span class="s">}</span> | |
228 215 <span class="s">}</span> | |
229 216 <span class="s">}</span> | |
230 217 <span class="k">if</span> <span class="s">(</span><span class="i">$NonNumericalDataFound</span><span class="s">)</span> <span class="s">{</span> | |
231 218 <span class="i">$PrintTextLine</span> = <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >= <span class="n">3</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
232 219 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} >=<span class="n">2</span> <span class="s">)</span> <span class="s">{</span> | |
233 220 <span class="k">print</span> <span class="q">"Line number $LineCount contains non-numerical data for some specified column(s)...\n"</span><span class="sc">;</span> | |
234 221 <span class="s">}</span> | |
235 222 <span class="s">}</span> | |
236 223 <span class="s">}</span> | |
237 224 <span class="k">if</span> <span class="s">(</span><span class="i">$PrintTextLine</span><span class="s">)</span> <span class="s">{</span> | |
238 225 <span class="k">print</span> <span class="q">"Line $LineCount: $Line\n\n"</span><span class="sc">;</span> | |
239 226 <span class="s">}</span> | |
240 227 <span class="s">}</span> | |
241 228 <span class="s">}</span> | |
242 229 <span class="s">}</span> | |
243 230 <span class="k">else</span> <span class="s">{</span> | |
244 231 <span class="k">while</span> <span class="s">(</span><span class="q"><TEXTFILE></span><span class="s">)</span> <span class="s">{</span> | |
245 232 <span class="i">$LineCount</span>++<span class="sc">;</span> | |
246 233 <span class="s">}</span> | |
247 234 <span class="s">}</span> | |
248 235 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span> | |
249 236 | |
250 237 <span class="k">print</span> <span class="q">"\nNumber of lines: $LineCount\n"</span><span class="sc">;</span> | |
251 238 <span class="k">print</span> <span class="q">"Number of columns: $TextFilesInfo{ColCount}[$Index]\n"</span><span class="sc">;</span> | |
252 239 <span class="k">print</span> <span class="q">"Column labels: "</span><span class="cm">,</span> <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="q">", "</span><span class="cm">,</span> <span class="n">1</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span> | |
253 240 | |
254 241 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CountEmpty</span>}<span class="s">)</span> <span class="s">{</span> | |
255 242 <span class="k">print</span> <span class="q">"\nNumber of lines with no data: $EmptyLinesCount\n"</span><span class="sc">;</span> | |
256 243 <span class="k">print</span> <span class="q">"Number of lines with some missing column data: $EmptyColDataLinesCount\n"</span><span class="sc">;</span> | |
257 244 <span class="k">print</span> <span class="q">"Number of lines containing greater than "</span><span class="cm">,</span> <span class="k">scalar</span><span class="s">(</span><span class="i">@ColLabels</span><span class="s">)</span><span class="cm">,</span> <span class="q">" columns: $GreaterThanMaxColLinesCount\n"</span><span class="sc">;</span> | |
258 245 <span class="i">PrintDataInformation</span><span class="s">(</span><span class="q">"Number of non-empty values for each column(s)"</span><span class="cm">,</span> \<span class="i">@ColLabels</span><span class="cm">,</span> \<span class="i">%NonEmptyColValuesCountMap</span><span class="s">)</span><span class="sc">;</span> | |
259 246 <span class="i">PrintDataInformation</span><span class="s">(</span><span class="q">"Number of empty values for each column(s)"</span><span class="cm">,</span> \<span class="i">@ColLabels</span><span class="cm">,</span> \<span class="i">%EmptyColValuesCountMap</span><span class="s">)</span><span class="sc">;</span> | |
260 247 <span class="s">}</span> | |
261 248 | |
262 249 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CheckData</span>}<span class="s">)</span> <span class="s">{</span> | |
263 250 <span class="k">print</span> <span class="q">"\n"</span><span class="sc">;</span> | |
264 251 <span class="i">PrintDataInformation</span><span class="s">(</span><span class="q">"Number of non-numerical data values for each column(s)"</span><span class="cm">,</span> \<span class="i">@ColLabels</span><span class="cm">,</span> \<span class="i">%NonNumericalColValuesCountMap</span><span class="s">)</span><span class="sc">;</span> | |
265 252 <span class="i">PrintDataInformation</span><span class="s">(</span><span class="q">"Number of numerical data values for each column(s)"</span><span class="cm">,</span> \<span class="i">@ColLabels</span><span class="cm">,</span> \<span class="i">%NumericalColValuesCountMap</span><span class="s">)</span><span class="sc">;</span> | |
266 253 <span class="k">print</span> <span class="q">"\n"</span><span class="sc">;</span> | |
267 254 <span class="s">}</span> | |
268 255 | |
269 256 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">CheckNumericalData</span>} && <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColLabels</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span> | |
270 257 <span class="i">PrintDataInformation</span><span class="s">(</span><span class="q">"Number of non-numerical data values for each column(s)"</span><span class="cm">,</span> \<span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColLabels</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> \<span class="i">%SpecifiedNonNumericalColValuesCountMap</span><span class="s">)</span><span class="sc">;</span> | |
271 258 <span class="s">}</span> | |
272 259 | |
273 260 <span class="c"># File size and modification information...</span> | |
274 261 <span class="k">print</span> <span class="q">"\nFile size: "</span><span class="cm">,</span> <span class="i">FormatFileSize</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>]<span class="s">)</span><span class="cm">,</span> <span class="q">" \n"</span><span class="sc">;</span> | |
275 262 <span class="k">print</span> <span class="q">"Last modified: "</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">FileLastModified</span>}[<span class="i">$Index</span>]<span class="cm">,</span> <span class="q">" \n"</span><span class="sc">;</span> | |
276 263 <span class="s">}</span> | |
277 264 | |
278 265 <span class="c"># Total size of all the fiels...</span> | |
279 <a name="ListTotalSizeOfFiles-"></a> 266 <span class="k">sub </span><span class="m">ListTotalSizeOfFiles</span> <span class="s">{</span> | |
280 267 <span class="k">my</span><span class="s">(</span><span class="i">$FileOkayCount</span><span class="cm">,</span> <span class="i">$TotalSize</span><span class="cm">,</span> <span class="i">$Index</span><span class="s">)</span><span class="sc">;</span> | |
281 268 | |
282 269 <span class="i">$FileOkayCount</span> = <span class="n">0</span><span class="sc">;</span> | |
283 270 <span class="i">$TotalSize</span> = <span class="n">0</span><span class="sc">;</span> | |
284 271 | |
285 272 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span> | |
286 273 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span> | |
287 274 <span class="i">$FileOkayCount</span>++<span class="sc">;</span> | |
288 275 <span class="i">$TotalSize</span> += <span class="i">$TextFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>]<span class="sc">;</span> | |
289 276 <span class="s">}</span> | |
290 277 <span class="s">}</span> | |
291 278 <span class="k">if</span> <span class="s">(</span><span class="i">$FileOkayCount</span> > <span class="n">1</span><span class="s">)</span> <span class="s">{</span> | |
292 279 <span class="k">print</span> <span class="q">"\nTotal size of $FileOkayCount files: "</span><span class="cm">,</span> <span class="i">FormatFileSize</span><span class="s">(</span><span class="i">$TotalSize</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span> | |
293 280 <span class="s">}</span> | |
294 281 <span class="s">}</span> | |
295 282 | |
296 283 <span class="c"># List data information...</span> | |
297 <a name="PrintDataInformation-"></a> 284 <span class="k">sub </span><span class="m">PrintDataInformation</span> <span class="s">{</span> | |
298 285 <span class="k">my</span><span class="s">(</span><span class="i">$InfoLabel</span><span class="cm">,</span> <span class="i">$DataLabelRef</span><span class="cm">,</span> <span class="i">$DataLabelToValueMapRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span> | |
299 286 <span class="k">my</span><span class="s">(</span><span class="i">$Line</span><span class="cm">,</span> <span class="i">$Label</span><span class="s">)</span><span class="sc">;</span> | |
300 287 | |
301 288 <span class="i">$Line</span> = <span class="q">""</span><span class="sc">;</span> | |
302 289 <span class="k">for</span> <span class="i">$Label</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$DataLabelRef</span>}<span class="s">)</span> <span class="s">{</span> | |
303 290 <span class="i">$Line</span> .= <span class="q">" \"$Label\" - "</span> . <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$DataLabelToValueMapRef</span>->{<span class="i">$Label</span>}<span class="s">)</span> ? <span class="i">$DataLabelToValueMapRef</span>->{<span class="i">$Label</span>} <span class="co">:</span> <span class="n">0</span><span class="s">)</span> . <span class="q">","</span><span class="sc">;</span> | |
304 291 <span class="s">}</span> | |
305 292 <span class="i">$Line</span> =~ <span class="q">s/\,$//g</span><span class="sc">;</span> | |
306 293 <span class="k">print</span> <span class="q">"$InfoLabel: $Line\n"</span><span class="sc">;</span> | |
307 294 <span class="s">}</span> | |
308 295 | |
309 296 <span class="c"># Retrieve information about input text files...</span> | |
310 <a name="RetrieveTextFilesInfo-"></a> 297 <span class="k">sub </span><span class="m">RetrieveTextFilesInfo</span> <span class="s">{</span> | |
311 298 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$TextFile</span><span class="cm">,</span> <span class="i">$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">$InDelim</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="cm">,</span> <span class="i">$ModifiedTimeString</span><span class="cm">,</span> <span class="i">$ModifiedDateString</span><span class="s">)</span><span class="sc">;</span> | |
312 299 | |
313 300 <span class="i">%TextFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
314 301 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
315 302 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
316 303 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
317 304 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
318 305 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
319 306 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">FileSize</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
320 307 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">FileLastModified</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
321 308 | |
322 309 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span> | |
323 310 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span> | |
324 311 | |
325 312 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span> | |
326 313 <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span> | |
327 314 <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span> | |
328 315 <span class="i">$TextFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span> | |
329 316 <span class="i">$TextFilesInfo</span>{<span class="w">FileLastModified</span>}[<span class="i">$Index</span>] = <span class="q">''</span><span class="sc">;</span> | |
330 317 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
331 318 <span class="i">%</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
332 319 | |
333 320 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$TextFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
334 321 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: It doesn't exist\n"</span><span class="sc">;</span> | |
335 322 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span> | |
336 323 <span class="s">}</span> | |
337 324 <span class="k">if</span> <span class="s">(</span>!<span class="i">CheckFileType</span><span class="s">(</span><span class="i">$TextFile</span><span class="cm">,</span> <span class="q">"csv tsv"</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
338 325 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: It's not a csv or tsv file\n"</span><span class="sc">;</span> | |
339 326 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span> | |
340 327 <span class="s">}</span> | |
341 328 <span class="s">(</span><span class="i">$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">$TextFile</span><span class="s">)</span><span class="sc">;</span> | |
342 329 <span class="k">if</span> <span class="s">(</span><span class="i">$FileExt</span> =~ <span class="q">/^tsv$/i</span><span class="s">)</span> <span class="s">{</span> | |
343 330 <span class="i">$InDelim</span> = <span class="q">"\t"</span><span class="sc">;</span> | |
344 331 <span class="s">}</span> | |
345 332 <span class="k">else</span> <span class="s">{</span> | |
346 333 <span class="i">$InDelim</span> = <span class="q">"\,"</span><span class="sc">;</span> | |
347 334 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span> | |
348 335 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The value specified, $OptionsInfo{InDelim}, for option \"--indelim\" is not valid for csv files\n"</span><span class="sc">;</span> | |
349 336 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span> | |
350 337 <span class="s">}</span> | |
351 338 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} =~ <span class="q">/^semicolon$/i</span><span class="s">)</span> <span class="s">{</span> | |
352 339 <span class="i">$InDelim</span> = <span class="q">"\;"</span><span class="sc">;</span> | |
353 340 <span class="s">}</span> | |
354 341 <span class="s">}</span> | |
355 342 | |
356 343 <span class="k">if</span> <span class="s">(</span>!<span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">"$TextFile"</span><span class="s">)</span> <span class="s">{</span> | |
357 344 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: Couldn't open it: $! \n"</span><span class="sc">;</span> | |
358 345 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span> | |
359 346 <span class="s">}</span> | |
360 347 | |
361 348 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="sc">;</span> | |
362 349 <span class="i">@ColLabels</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span> | |
363 350 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span> | |
364 351 | |
365 352 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span> | |
366 353 <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="i">$InDelim</span><span class="sc">;</span> | |
367 354 | |
368 355 <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] = <span class="i">@ColLabels</span><span class="sc">;</span> | |
369 356 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@ColLabels</span><span class="sc">;</span> | |
370 357 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#ColLabels</span><span class="s">)</span> <span class="s">{</span> | |
371 358 <span class="i">$ColLabel</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span> | |
372 359 <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>} = <span class="i">$ColNum</span><span class="sc">;</span> | |
373 360 <span class="s">}</span> | |
374 361 <span class="i">$TextFilesInfo</span>{<span class="w">FileSize</span>}[<span class="i">$Index</span>] = <span class="i">FileSize</span><span class="s">(</span><span class="i">$TextFile</span><span class="s">)</span><span class="sc">;</span> | |
375 362 <span class="s">(</span><span class="i">$ModifiedTimeString</span><span class="cm">,</span> <span class="i">$ModifiedDateString</span><span class="s">)</span> = <span class="i">FormattedFileModificationTimeAndDate</span><span class="s">(</span><span class="i">$TextFile</span><span class="s">)</span><span class="sc">;</span> | |
376 363 <span class="i">$TextFilesInfo</span>{<span class="w">FileLastModified</span>}[<span class="i">$Index</span>] = <span class="q">"$ModifiedTimeString; $ModifiedDateString"</span><span class="sc">;</span> | |
377 364 <span class="s">}</span> | |
378 365 | |
379 366 <span class="s">}</span> | |
380 367 | |
381 368 <span class="c"># Make sure specified numerical data columns are okay...</span> | |
382 <a name="ProcessColumnsInfo-"></a> 369 <span class="k">sub </span><span class="m">ProcessColumnsInfo</span> <span class="s">{</span> | |
383 370 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$TextFile</span><span class="s">)</span><span class="sc">;</span> | |
384 371 | |
385 372 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
386 373 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
387 374 | |
388 375 <span class="j">FILELIST:</span> <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span> | |
389 376 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span> | |
390 377 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColNums</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
391 378 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColLabels</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
392 379 | |
393 380 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span> | |
394 381 <span class="k">my</span><span class="s">(</span><span class="i">$SpecifiedColNum</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="cm">,</span> <span class="i">@SpecifiedColNums</span><span class="cm">,</span> <span class="i">@SpecifiedColLabels</span><span class="s">)</span><span class="sc">;</span> | |
395 382 <span class="i">@SpecifiedColNums</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
396 383 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span> | |
397 384 <span class="k">for</span> <span class="i">$SpecifiedColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedNumericalDataCols</span>}}<span class="s">)</span> <span class="s">{</span> | |
398 385 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedColNum</span> <= <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span> | |
399 386 <span class="i">$ColNum</span> = <span class="i">$SpecifiedColNum</span> - <span class="n">1</span><span class="sc">;</span> | |
400 387 <span class="k">push</span> <span class="i">@SpecifiedColNums</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span> | |
401 388 <span class="k">push</span> <span class="i">@SpecifiedColLabels</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>][<span class="i">$ColNum</span>]<span class="sc">;</span> | |
402 389 <span class="s">}</span> | |
403 390 <span class="s">}</span> | |
404 391 <span class="s">}</span> | |
405 392 <span class="k">else</span> <span class="s">{</span> | |
406 393 <span class="k">for</span> <span class="i">$ColLabel</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedNumericalDataCols</span>}}<span class="s">)</span> <span class="s">{</span> | |
407 394 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
408 395 <span class="i">$ColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>}<span class="sc">;</span> | |
409 396 <span class="k">push</span> <span class="i">@SpecifiedColNums</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span> | |
410 397 <span class="k">push</span> <span class="i">@SpecifiedColLabels</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="sc">;</span> | |
411 398 <span class="s">}</span> | |
412 399 <span class="s">}</span> | |
413 400 <span class="s">}</span> | |
414 401 <span class="k">if</span> <span class="s">(</span><span class="i">@SpecifiedColNums</span><span class="s">)</span> <span class="s">{</span> | |
415 402 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColNums</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@SpecifiedColNums</span><span class="sc">;</span> | |
416 403 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">NumericalDataColLabels</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@SpecifiedColLabels</span><span class="sc">;</span> | |
417 404 <span class="s">}</span> | |
418 405 <span class="s">}</span> | |
419 406 <span class="s">}</span> | |
420 407 <span class="s">}</span> | |
421 408 | |
422 409 <span class="c"># Process option values...</span> | |
423 <a name="ProcessOptions-"></a> 410 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span> | |
424 411 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
425 412 | |
426 413 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span> | |
427 414 | |
428 415 <span class="i">$OptionsInfo</span>{<span class="w">All</span>} = <span class="i">$Options</span>{<span class="w">all</span>} ? <span class="i">$Options</span>{<span class="w">all</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
429 416 <span class="i">$OptionsInfo</span>{<span class="w">Count</span>} = <span class="i">$Options</span>{<span class="w">count</span>} ? <span class="i">$Options</span>{<span class="w">count</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
430 417 | |
431 418 <span class="i">$OptionsInfo</span>{<span class="w">DetailLevel</span>} = <span class="i">$Options</span>{<span class="w">detail</span>} ? <span class="i">$Options</span>{<span class="w">detail</span>} <span class="co">:</span> <span class="n">1</span><span class="sc">;</span> | |
432 419 | |
433 420 <span class="i">$OptionsInfo</span>{<span class="w">Empty</span>} = <span class="i">$Options</span>{<span class="w">empty</span>} ? <span class="i">$Options</span>{<span class="w">empty</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
434 421 | |
435 422 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="i">$Options</span>{<span class="w">indelim</span>}<span class="sc">;</span> | |
436 423 <span class="i">$OptionsInfo</span>{<span class="w">NumericalDataCols</span>} = <span class="i">$Options</span>{<span class="w">numericaldatacols</span>} ? <span class="i">$Options</span>{<span class="w">numericaldatacols</span>} <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
437 424 | |
438 425 <span class="i">$OptionsInfo</span>{<span class="w">ParseLines</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">empty</span>} || <span class="i">$Options</span>{<span class="w">numericaldatacols</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
439 426 <span class="i">$OptionsInfo</span>{<span class="w">CountEmpty</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">empty</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
440 427 <span class="i">$OptionsInfo</span>{<span class="w">CheckData</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">datacheck</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
441 428 <span class="i">$OptionsInfo</span>{<span class="w">CheckNumericalData</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">all</span>} || <span class="i">$Options</span>{<span class="w">numericaldatacols</span>}<span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span> | |
442 429 | |
443 430 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedNumericalDataCols</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
444 431 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">numericaldatacols</span>}<span class="s">)</span> <span class="s">{</span> | |
445 432 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedNumericalDataCols</span>}} = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">numericaldatacols</span>}<span class="sc">;</span> | |
446 433 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span> | |
447 434 <span class="k">my</span><span class="s">(</span><span class="i">$ColNum</span><span class="s">)</span><span class="sc">;</span> | |
448 435 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">SpecifiedNumericalDataCols</span>}}<span class="s">)</span> <span class="s">{</span> | |
449 436 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$ColNum</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
450 437 <span class="k">die</span> <span class="q">"Error: Invalid value $ColNum specified using \"--numericaldatacols\" option: Allowed values: > 0\n"</span><span class="sc">;</span> | |
451 438 <span class="s">}</span> | |
452 439 <span class="s">}</span> | |
453 440 <span class="s">}</span> | |
454 441 <span class="s">}</span> | |
455 442 | |
456 443 <span class="s">}</span> | |
457 444 | |
458 445 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span> | |
459 <a name="SetupScriptUsage-"></a> 446 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span> | |
460 447 | |
461 448 <span class="c"># Retrieve all the options...</span> | |
462 449 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span> | |
463 450 <span class="i">$Options</span>{<span class="w">detail</span>} = <span class="n">1</span><span class="sc">;</span> | |
464 451 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">"colnum"</span><span class="sc">;</span> | |
465 452 <span class="i">$Options</span>{<span class="w">indelim</span>} = <span class="q">"comma"</span><span class="sc">;</span> | |
466 453 <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">"all|a"</span><span class="cm">,</span> <span class="q">"count|c"</span><span class="cm">,</span> <span class="q">"datacheck"</span><span class="cm">,</span> <span class="q">"detail|d=i"</span><span class="cm">,</span> <span class="q">"empty|e"</span><span class="cm">,</span> <span class="q">"help|h"</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">"numericaldatacols|n=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> | |
467 454 <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 $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span> | |
468 455 <span class="s">}</span> | |
469 456 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span> | |
470 457 <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> | |
471 458 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span> | |
472 459 <span class="s">}</span> | |
473 460 <span class="k">chdir</span> <span class="i">$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}: $! \n"</span><span class="sc">;</span> | |
474 461 <span class="s">}</span> | |
475 462 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(colnum|collabel)$/i</span><span class="s">)</span> <span class="s">{</span> | |
476 463 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: colnum or collabel\n"</span><span class="sc">;</span> | |
477 464 <span class="s">}</span> | |
478 465 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span> | |
479 466 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n"</span><span class="sc">;</span> | |
480 467 <span class="s">}</span> | |
481 468 <span class="k">if</span> <span class="s">(</span>!<span class="i">IsPositiveInteger</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">detail</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span> | |
482 469 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{detail}, for option \"-d --detail\" is not valid. Allowed values: > 0\n"</span><span class="sc">;</span> | |
483 470 <span class="s">}</span> | |
484 471 <span class="s">}</span> | |
485 472 | |
486 <a name="EOF-"></a></pre> | |
487 <p> </p> | |
488 <br /> | |
489 <center> | |
490 <img src="../../../images/h2o2.png"> | |
491 </center> | |
492 </body> | |
493 </html> |