0
|
1 <html>
|
|
2 <head>
|
|
3 <title>MayaChemTools:Code:MergeTextFiles.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: MergeTextFiles.pl,v $</span>
|
|
17 4 <span class="c"># $Date: 2015/02/28 20:46:20 $</span>
|
|
18 5 <span class="c"># $Revision: 1.40 $</span>
|
|
19 6 <span class="c">#</span>
|
|
20 7 <span class="c"># Author: Manish Sud <msud@san.rr.com></span>
|
|
21 8 <span class="c">#</span>
|
|
22 9 <span class="c"># Copyright (C) 2015 Manish Sud. All rights reserved.</span>
|
|
23 10 <span class="c">#</span>
|
|
24 11 <span class="c"># This file is part of MayaChemTools.</span>
|
|
25 12 <span class="c">#</span>
|
|
26 13 <span class="c"># MayaChemTools is free software; you can redistribute it and/or modify it under</span>
|
|
27 14 <span class="c"># the terms of the GNU Lesser General Public License as published by the Free</span>
|
|
28 15 <span class="c"># Software Foundation; either version 3 of the License, or (at your option) any</span>
|
|
29 16 <span class="c"># later version.</span>
|
|
30 17 <span class="c">#</span>
|
|
31 18 <span class="c"># MayaChemTools is distributed in the hope that it will be useful, but without</span>
|
|
32 19 <span class="c"># any warranty; without even the implied warranty of merchantability of fitness</span>
|
|
33 20 <span class="c"># for a particular purpose. See the GNU Lesser General Public License for more</span>
|
|
34 21 <span class="c"># details.</span>
|
|
35 22 <span class="c">#</span>
|
|
36 23 <span class="c"># You should have received a copy of the GNU Lesser General Public License</span>
|
|
37 24 <span class="c"># along with MayaChemTools; if not, see <http://www.gnu.org/licenses/> or</span>
|
|
38 25 <span class="c"># write to the Free Software Foundation Inc., 59 Temple Place, Suite 330,</span>
|
|
39 26 <span class="c"># Boston, MA, 02111-1307, USA.</span>
|
|
40 27 <span class="c">#</span>
|
|
41 28
|
|
42 29 <span class="k">use</span> <span class="w">strict</span><span class="sc">;</span>
|
|
43 30 <span class="k">use</span> <span class="w">FindBin</span><span class="sc">;</span> <span class="k">use</span> <span class="w">lib</span> <span class="q">"$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">FileHandle</span><span class="sc">;</span>
|
|
49 36 <span class="k">use</span> <span class="w">FileUtil</span><span class="sc">;</span>
|
|
50 37 <span class="k">use</span> <span class="w">TextUtil</span><span class="sc">;</span>
|
|
51 38
|
|
52 39 <span class="k">my</span><span class="s">(</span><span class="i">$ScriptName</span><span class="cm">,</span> <span class="i">%Options</span><span class="cm">,</span> <span class="i">$StartTime</span><span class="cm">,</span> <span class="i">$EndTime</span><span class="cm">,</span> <span class="i">$TotalTime</span><span class="s">)</span><span class="sc">;</span>
|
|
53 40
|
|
54 41 <span class="c"># Autoflush STDOUT</span>
|
|
55 42 <span class="i">$|</span> = <span class="n">1</span><span class="sc">;</span>
|
|
56 43
|
|
57 44 <span class="c"># Starting message...</span>
|
|
58 45 <span class="i">$ScriptName</span> = <span class="w">basename</span> <span class="i">$0</span><span class="sc">;</span>
|
|
59 46 <span class="k">print</span> <span class="q">"\n$ScriptName:Starting...\n\n"</span><span class="sc">;</span>
|
|
60 47 <span class="i">$StartTime</span> = <span class="w">new</span> <span class="w">Benchmark</span><span class="sc">;</span>
|
|
61 48
|
|
62 49 <span class="c"># Get the options and setup script...</span>
|
|
63 50 <span class="i">SetupScriptUsage</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
64 51 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">help</span>} || <span class="i">@ARGV</span> < <span class="n">1</span><span class="s">)</span> <span class="s">{</span>
|
|
65 52 <span class="k">die</span> <span class="i">GetUsageFromPod</span><span class="s">(</span><span class="q">"$FindBin::Bin/$ScriptName"</span><span class="s">)</span><span class="sc">;</span>
|
|
66 53 <span class="s">}</span>
|
|
67 54
|
|
68 55 <span class="k">my</span><span class="s">(</span><span class="i">@TextFilesList</span><span class="s">)</span><span class="sc">;</span>
|
|
69 56 <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>
|
|
70 57
|
|
71 58 <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>
|
|
72 59 <span class="k">die</span> <span class="q">"Error: Specify more than one text file.\n"</span><span class="sc">;</span>
|
|
73 60 <span class="s">}</span>
|
|
74 61
|
|
75 62 <span class="c"># Process options...</span>
|
|
76 63 <span class="k">print</span> <span class="q">"Processing options...\n"</span><span class="sc">;</span>
|
|
77 64 <span class="k">my</span><span class="s">(</span><span class="i">%OptionsInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
78 65 <span class="i">ProcessOptions</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
79 66
|
|
80 67 <span class="c"># Setup information about input files...</span>
|
|
81 68 <span class="k">my</span><span class="s">(</span><span class="i">%TextFilesInfo</span><span class="s">)</span><span class="sc">;</span>
|
|
82 69 <span class="k">print</span> <span class="q">"Checking input text files...\n"</span><span class="sc">;</span>
|
|
83 70 <span class="i">RetrieveTextFilesInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
84 71 <span class="i">RetrieveColumnsAndKeysInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
85 72
|
|
86 73 <span class="c"># Merge files...</span>
|
|
87 74 <span class="k">print</span> <span class="q">"\nGenerating new text file $OptionsInfo{NewTextFile}...\n"</span><span class="sc">;</span>
|
|
88 75 <span class="i">MergeTextFiles</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
89 76
|
|
90 77 <span class="k">print</span> <span class="q">"\n$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">$TotalTime</span><span class="s">)</span><span class="cm">,</span> <span class="q">"\n"</span><span class="sc">;</span>
|
|
95 82
|
|
96 83 <span class="c">###############################################################################</span>
|
|
97 84
|
|
98 85 <span class="c"># Merge all valid Text files...</span>
|
|
99 <a name="MergeTextFiles-"></a> 86 <span class="k">sub </span><span class="m">MergeTextFiles</span> <span class="s">{</span>
|
|
100 87 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="s">)</span><span class="sc">;</span>
|
|
101 88
|
|
102 89 <span class="k">open</span> <span class="w">NEWTEXTFILE</span><span class="cm">,</span> <span class="q">">$OptionsInfo{NewTextFile}"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $OptionsInfo{NewTextFile}: $! \n"</span><span class="sc">;</span>
|
|
103 90
|
|
104 91 <span class="i">WriteNewTextFileColumnLabels</span><span class="s">(</span>\<span class="i">*NEWTEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
105 92
|
|
106 93 <span class="c">#Open up all the files and skip coumn label line...</span>
|
|
107 94 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
108 95 <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>
|
|
109 96 <span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="i">$Index</span>] = <span class="w">new</span> <span class="w">FileHandle</span><span class="sc">;</span>
|
|
110 97
|
|
111 98 <span class="k">open</span> <span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="i">$Index</span>]<span class="cm">,</span> <span class="q">"$TextFilesList[$Index]"</span> <span class="k">or</span> <span class="k">die</span> <span class="q">"Error: Couldn't open $TextFilesList[$Index]: $! \n"</span><span class="sc">;</span>
|
|
112 99 <span class="i">GetTextLine</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="i">$Index</span>]<span class="s">)</span><span class="sc">;</span>
|
|
113 100 <span class="s">}</span>
|
|
114 101
|
|
115 102 <span class="c"># Merge files...</span>
|
|
116 103 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Keys</span>}<span class="s">)</span> <span class="s">{</span>
|
|
117 104 <span class="i">MergeColumnValuesUsingKeys</span><span class="s">(</span>\<span class="i">*NEWTEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
118 105 <span class="s">}</span>
|
|
119 106 <span class="k">else</span> <span class="s">{</span>
|
|
120 107 <span class="i">MergeColumnValues</span><span class="s">(</span>\<span class="i">*NEWTEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
121 108 <span class="s">}</span>
|
|
122 109
|
|
123 110 <span class="c"># Close all opened files...</span>
|
|
124 111 <span class="k">close</span> <span class="w">NEWTEXTFILE</span><span class="sc">;</span>
|
|
125 112 <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>
|
|
126 113 <span class="k">close</span> <span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
127 114 <span class="s">}</span>
|
|
128 115
|
|
129 116 <span class="s">}</span>
|
|
130 117
|
|
131 118 <span class="c"># Merge all the column values...</span>
|
|
132 <a name="MergeColumnValues-"></a> 119 <span class="k">sub </span><span class="m">MergeColumnValues</span> <span class="s">{</span>
|
|
133 120 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
134 121 <span class="k">my</span><span class="s">(</span><span class="i">$Index</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">$Value</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@File1LineWords</span><span class="cm">,</span> <span class="i">@ColValues</span><span class="s">)</span><span class="sc">;</span>
|
|
135 122
|
|
136 123 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="n">0</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
137 124 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
138 125 <span class="i">@ColValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
139 126
|
|
140 127 <span class="c">#Collect column values from first file before the merge point...</span>
|
|
141 128 <span class="i">@File1LineWords</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>
|
|
142 129 <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">File1Part1ColNums</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
143 130 <span class="i">$Value</span> = <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@File1LineWords</span><span class="s">)</span> ? <span class="i">$File1LineWords</span>[<span class="i">$ColNum</span>] <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
144 131 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
145 132 <span class="s">}</span>
|
|
146 133
|
|
147 134 <span class="c">#Collect column values from other text files...</span>
|
|
148 135 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
149 136 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
150 137 <span class="k">if</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="i">$Index</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
151 138 <span class="i">@LineWords</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
152 139 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
153 140 <span class="i">$Value</span> = <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@LineWords</span><span class="s">)</span> ? <span class="i">$LineWords</span>[<span class="i">$ColNum</span>] <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
154 141 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
155 142 <span class="s">}</span>
|
|
156 143 <span class="s">}</span>
|
|
157 144 <span class="s">}</span>
|
|
158 145
|
|
159 146 <span class="c">#Collect column labels from first file after the merge point...</span>
|
|
160 147 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part2ColNums</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
161 148 <span class="i">$Value</span> = <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@File1LineWords</span><span class="s">)</span> ? <span class="i">$File1LineWords</span>[<span class="i">$ColNum</span>] <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
162 149 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
163 150 <span class="s">}</span>
|
|
164 151
|
|
165 152 <span class="c"># Write it out...</span>
|
|
166 153 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
167 154 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
168 155 <span class="s">}</span>
|
|
169 156
|
|
170 157 <span class="s">}</span>
|
|
171 158
|
|
172 159 <span class="c"># Merge column values using keys...</span>
|
|
173 <a name="MergeColumnValuesUsingKeys-"></a> 160 <span class="k">sub </span><span class="m">MergeColumnValuesUsingKeys</span> <span class="s">{</span>
|
|
174 161 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
175 162 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$InDelim</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$Value</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$KeyColNum</span><span class="cm">,</span> <span class="i">$KeyColValue</span><span class="cm">,</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">@File1LineWords</span><span class="cm">,</span> <span class="i">@TextFilesKeysToLinesMap</span><span class="s">)</span><span class="sc">;</span>
|
|
176 163
|
|
177 164 <span class="i">@TextFilesKeysToLinesMap</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
178 165
|
|
179 166 <span class="c"># Retrieve text lines from all the files except for the first file...</span>
|
|
180 167 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
181 168 <span class="i">%</span>{<span class="i">$TextFilesKeysToLinesMap</span>[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
182 169
|
|
183 170 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
184 171 <span class="i">$KeyColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
185 172
|
|
186 173 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="i">$Index</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
187 174 <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>
|
|
188 175 <span class="k">if</span> <span class="s">(</span><span class="i">$KeyColNum</span> < <span class="i">@LineWords</span><span class="s">)</span> <span class="s">{</span>
|
|
189 176 <span class="i">$KeyColValue</span> = <span class="i">$LineWords</span>[<span class="i">$KeyColNum</span>]<span class="sc">;</span>
|
|
190 177 <span class="k">if</span> <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$KeyColValue</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
191 178 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesKeysToLinesMap</span>[<span class="i">$Index</span>]{<span class="i">$KeyColValue</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
192 179 <span class="k">warn</span> <span class="q">"Warning: Ignoring line, $Line, in text file $TextFilesList[$Index]: Column key value, $KeyColValue, already exists\n"</span><span class="sc">;</span>
|
|
193 180 <span class="s">}</span>
|
|
194 181 <span class="k">else</span> <span class="s">{</span>
|
|
195 182 <span class="i">@</span>{<span class="i">$TextFilesKeysToLinesMap</span>[<span class="i">$Index</span>]{<span class="i">$KeyColValue</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
196 183 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesKeysToLinesMap</span>[<span class="i">$Index</span>]{<span class="i">$KeyColValue</span>}}<span class="cm">,</span> <span class="i">@LineWords</span><span class="sc">;</span>
|
|
197 184 <span class="s">}</span>
|
|
198 185 <span class="s">}</span>
|
|
199 186 <span class="s">}</span>
|
|
200 187 <span class="s">}</span>
|
|
201 188 <span class="s">}</span>
|
|
202 189
|
|
203 190 <span class="k">while</span> <span class="s">(</span><span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">FileHandle</span>}[<span class="n">0</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
204 191 <span class="i">$InDelim</span> = <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
205 192
|
|
206 193 <span class="i">@ColValues</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
207 194 <span class="i">@File1LineWords</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
208 195
|
|
209 196 <span class="i">$KeyColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}[<span class="n">0</span>]<span class="sc">;</span>
|
|
210 197 <span class="i">$KeyColValue</span> = <span class="i">$File1LineWords</span>[<span class="i">$KeyColNum</span>]<span class="sc">;</span>
|
|
211 198
|
|
212 199 <span class="c">#Collect column values from first file before the merge point...</span>
|
|
213 200 <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">File1Part1ColNums</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
214 201 <span class="i">$Value</span> = <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@File1LineWords</span><span class="s">)</span> ? <span class="i">$File1LineWords</span>[<span class="i">$ColNum</span>] <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
215 202 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
216 203 <span class="s">}</span>
|
|
217 204
|
|
218 205 <span class="c">#Collect column values from other text files...</span>
|
|
219 206 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
220 207 <span class="i">@LineWords</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
221 208 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesKeysToLinesMap</span>[<span class="i">$Index</span>]{<span class="i">$KeyColValue</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
222 209 <span class="k">push</span> <span class="i">@LineWords</span><span class="cm">,</span> <span class="i">@</span>{<span class="i">$TextFilesKeysToLinesMap</span>[<span class="i">$Index</span>]{<span class="i">$KeyColValue</span>}}<span class="sc">;</span>
|
|
223 210 <span class="s">}</span>
|
|
224 211 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
225 212 <span class="i">$Value</span> = <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@LineWords</span><span class="s">)</span> ? <span class="i">$LineWords</span>[<span class="i">$ColNum</span>] <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
226 213 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
227 214 <span class="s">}</span>
|
|
228 215 <span class="s">}</span>
|
|
229 216
|
|
230 217 <span class="c">#Collect column labels from first file after the merge point...</span>
|
|
231 218 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part2ColNums</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
232 219 <span class="i">$Value</span> = <span class="s">(</span><span class="i">$ColNum</span> < <span class="i">@File1LineWords</span><span class="s">)</span> ? <span class="i">$File1LineWords</span>[<span class="i">$ColNum</span>] <span class="co">:</span> <span class="q">""</span><span class="sc">;</span>
|
|
233 220 <span class="k">push</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$Value</span><span class="sc">;</span>
|
|
234 221 <span class="s">}</span>
|
|
235 222
|
|
236 223 <span class="c"># Write it out...</span>
|
|
237 224 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColValues</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
238 225 <span class="k">print</span> <span class="i">$NewTextFileRef</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
239 226 <span class="s">}</span>
|
|
240 227
|
|
241 228 <span class="s">}</span>
|
|
242 229
|
|
243 230 <span class="c"># Write out column labels for new merged text file...</span>
|
|
244 <a name="WriteNewTextFileColumnLabels-"></a> 231 <span class="k">sub </span><span class="m">WriteNewTextFileColumnLabels</span> <span class="s">{</span>
|
|
245 232 <span class="k">my</span><span class="s">(</span><span class="i">$NewTextFileRef</span><span class="s">)</span> = <span class="i">@_</span><span class="sc">;</span>
|
|
246 233 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Line</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="s">)</span><span class="sc">;</span>
|
|
247 234
|
|
248 235 <span class="c">#Write out column labels for the merged text file...</span>
|
|
249 236 <span class="i">@ColLabels</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
250 237
|
|
251 238 <span class="c">#Collect column labels from first file before the merge point...</span>
|
|
252 239 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part1ColNums</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
253 240 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="n">0</span>]{<span class="i">$ColNum</span>}<span class="sc">;</span>
|
|
254 241 <span class="s">}</span>
|
|
255 242
|
|
256 243 <span class="c">#Collect column labels from other text files...</span>
|
|
257 244 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
258 245 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
259 246 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColNum</span>}<span class="sc">;</span>
|
|
260 247 <span class="s">}</span>
|
|
261 248 <span class="s">}</span>
|
|
262 249
|
|
263 250 <span class="c">#Collect column labels from first file after the merge point...</span>
|
|
264 251 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part2ColNums</span>}}<span class="s">)</span> <span class="s">{</span>
|
|
265 252 <span class="k">push</span> <span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="n">0</span>]{<span class="i">$ColNum</span>}<span class="sc">;</span>
|
|
266 253 <span class="s">}</span>
|
|
267 254
|
|
268 255 <span class="c">#Write it out...</span>
|
|
269 256 <span class="i">$Line</span> = <span class="i">JoinWords</span><span class="s">(</span>\<span class="i">@ColLabels</span><span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>}<span class="cm">,</span> <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>}<span class="s">)</span><span class="sc">;</span>
|
|
270 257 <span class="k">print</span> <span class="i">NEWTEXTFILE</span> <span class="q">"$Line\n"</span><span class="sc">;</span>
|
|
271 258 <span class="s">}</span>
|
|
272 259
|
|
273 260 <span class="c"># Retrieve text file columns and keys information for specified options...</span>
|
|
274 <a name="RetrieveColumnsAndKeysInfo-"></a> 261 <span class="k">sub </span><span class="m">RetrieveColumnsAndKeysInfo</span> <span class="s">{</span>
|
|
275 262 <span class="i">ProcessColumnsInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
276 263
|
|
277 264 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Keys</span>}<span class="s">)</span> <span class="s">{</span>
|
|
278 265 <span class="i">ProcessKeysInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
279 266 <span class="s">}</span>
|
|
280 267
|
|
281 268 <span class="i">ProcessStartColInfo</span><span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
282 269 <span class="s">}</span>
|
|
283 270
|
|
284 271 <span class="c"># Process specified columns...</span>
|
|
285 <a name="ProcessColumnsInfo-"></a> 272 <span class="k">sub </span><span class="m">ProcessColumnsInfo</span> <span class="s">{</span>
|
|
286 273 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$SpecifiedColNum</span><span class="cm">,</span> <span class="i">$Values</span><span class="cm">,</span> <span class="i">$ColIndex</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">@Words</span><span class="s">)</span><span class="sc">;</span>
|
|
287 274
|
|
288 275 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
289 276 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
290 277 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
291 278
|
|
292 279 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
293 280
|
|
294 281 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
295 282
|
|
296 283 <span class="i">$Values</span> = <span class="q">"all"</span><span class="sc">;</span>
|
|
297 284 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Columns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
298 285 <span class="i">$Values</span> = <span class="i">$OptionsInfo</span>{<span class="w">ColValues</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
299 286 <span class="s">}</span>
|
|
300 287
|
|
301 288 <span class="k">if</span> <span class="s">(</span><span class="i">$Values</span> =~ <span class="q">/all/i</span><span class="s">)</span> <span class="s">{</span>
|
|
302 289 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^colnum$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
303 290 <span class="k">for</span> <span class="i">$ColNum</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
|
|
304 291 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
305 292 <span class="s">}</span>
|
|
306 293 <span class="s">}</span>
|
|
307 294 <span class="k">else</span> <span class="s">{</span>
|
|
308 295 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
309 296 <span class="s">}</span>
|
|
310 297 <span class="s">}</span>
|
|
311 298 <span class="k">else</span> <span class="s">{</span>
|
|
312 299 <span class="i">@Words</span> = <span class="k">split</span> <span class="q">","</span><span class="cm">,</span> <span class="i">$Values</span><span class="sc">;</span>
|
|
313 300 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@Words</span><span class="sc">;</span>
|
|
314 301 <span class="s">}</span>
|
|
315 302
|
|
316 303 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
317 304 <span class="i">%</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
318 305
|
|
319 306 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^collabel$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
320 307 <span class="k">for</span> <span class="i">$ColIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
321 308 <span class="i">$ColLabel</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>][<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
322 309 <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">$ColLabel</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
323 310 <span class="i">$ColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>}<span class="sc">;</span>
|
|
324 311 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
325 312 <span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColNum</span>} = <span class="i">$ColLabel</span><span class="sc">;</span>
|
|
326 313 <span class="s">}</span>
|
|
327 314 <span class="k">else</span> <span class="s">{</span>
|
|
328 315 <span class="k">warn</span> <span class="q">"Warning: Ignoring value, $ColLabel, specified using \"-c --column\" option: column name doesn't exist in $TextFilesList[$Index] \n"</span><span class="sc">;</span>
|
|
329 316 <span class="s">}</span>
|
|
330 317 <span class="s">}</span>
|
|
331 318 <span class="s">}</span>
|
|
332 319 <span class="k">else</span> <span class="s">{</span>
|
|
333 320 <span class="k">for</span> <span class="i">$ColIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
334 321 <span class="i">$SpecifiedColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColSpecified</span>}[<span class="i">$Index</span>][<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
335 322 <span class="k">if</span> <span class="s">(</span><span class="i">$SpecifiedColNum</span> > <span class="n">0</span> && <span class="i">$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>
|
|
336 323 <span class="i">$ColNum</span> = <span class="i">$SpecifiedColNum</span> - <span class="n">1</span><span class="sc">;</span>
|
|
337 324 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
338 325 <span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColNum</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>
|
|
339 326 <span class="s">}</span>
|
|
340 327 <span class="k">else</span> <span class="s">{</span>
|
|
341 328 <span class="k">warn</span> <span class="q">"Warning: Ignoring value, $SpecifiedColNum, specified using \"-c --column\" option: column number doesn't exist in $TextFilesList[$Index] \n"</span><span class="sc">;</span>
|
|
342 329 <span class="s">}</span>
|
|
343 330 <span class="s">}</span>
|
|
344 331 <span class="s">}</span>
|
|
345 332 <span class="k">my</span> <span class="s">(</span><span class="i">@ColToMergeSorted</span><span class="s">)</span> = <span class="k">sort</span> <span class="s">{</span> <span class="i">$a</span> <=> <span class="i">$b</span> <span class="s">}</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="sc">;</span>
|
|
346 333 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
347 334 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@ColToMergeSorted</span><span class="sc">;</span>
|
|
348 335 <span class="s">}</span>
|
|
349 336 <span class="s">}</span>
|
|
350 337
|
|
351 338 <span class="c"># Process specified key column values...</span>
|
|
352 <a name="ProcessKeysInfo-"></a> 339 <span class="k">sub </span><span class="m">ProcessKeysInfo</span> <span class="s">{</span>
|
|
353 340 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$Key</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="s">)</span><span class="sc">;</span>
|
|
354 341
|
|
355 342 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">KeysSpecified</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
356 343 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
357 344
|
|
358 345 <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>
|
|
359 346 <span class="i">$Key</span> = <span class="i">$OptionsInfo</span>{<span class="w">KeyValues</span>}[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
360 347
|
|
361 348 <span class="i">$TextFilesInfo</span>{<span class="w">KeysSpecified</span>}[<span class="i">$Index</span>] = <span class="i">$Key</span><span class="sc">;</span>
|
|
362 349 <span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}[<span class="i">$Index</span>] = <span class="n">-1</span><span class="sc">;</span>
|
|
363 350
|
|
364 351 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^collabel$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
365 352 <span class="i">$ColLabel</span> = <span class="i">$Key</span><span class="sc">;</span>
|
|
366 353 <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">$ColLabel</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
367 354 <span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}[<span class="i">$Index</span>] = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>}<span class="sc">;</span>
|
|
368 355 <span class="s">}</span>
|
|
369 356 <span class="k">else</span> <span class="s">{</span>
|
|
370 357 <span class="k">warn</span> <span class="q">"Warning: Ignoring value, $ColLabel, specified using \"-k --keys\" option: column name doesn't exist in $TextFilesList[$Index] \n"</span><span class="sc">;</span>
|
|
371 358 <span class="s">}</span>
|
|
372 359 <span class="s">}</span>
|
|
373 360 <span class="k">else</span> <span class="s">{</span>
|
|
374 361 <span class="i">$ColNum</span> = <span class="i">$Key</span><span class="sc">;</span>
|
|
375 362 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> > <span class="n">0</span> && <span class="i">$ColNum</span> <= <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>]<span class="s">)</span> <span class="s">{</span>
|
|
376 363 <span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}[<span class="i">$Index</span>] = <span class="i">$ColNum</span> - <span class="n">1</span><span class="sc">;</span>
|
|
377 364 <span class="s">}</span>
|
|
378 365 <span class="k">else</span> <span class="s">{</span>
|
|
379 366 <span class="k">warn</span> <span class="q">"Warning: Ignoring value, $ColNum, specified using \"-k --keys\" option: column number doesn't exist in $TextFilesList[$Index] \n"</span><span class="sc">;</span>
|
|
380 367 <span class="s">}</span>
|
|
381 368 <span class="s">}</span>
|
|
382 369 <span class="s">}</span>
|
|
383 370
|
|
384 371 <span class="c"># Modify columns to merge list to make sure the columns identified by key are taken off the list</span>
|
|
385 372 <span class="c"># except for the first text file...</span>
|
|
386 373 <span class="k">my</span><span class="s">(</span><span class="i">@ColToMergeFiltered</span><span class="s">)</span><span class="sc">;</span>
|
|
387 374
|
|
388 375 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">1</span> .. <span class="i">$#TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
389 376 <span class="i">@ColToMergeFiltered</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
390 377 <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">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
391 378 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">KeysToUse</span>}[<span class="i">$Index</span>] != <span class="i">$ColNum</span><span class="s">)</span> <span class="s">{</span>
|
|
392 379 <span class="k">push</span> <span class="i">@ColToMergeFiltered</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
393 380 <span class="s">}</span>
|
|
394 381 <span class="s">}</span>
|
|
395 382 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
396 383 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="i">$Index</span>]}<span class="cm">,</span> <span class="i">@ColToMergeFiltered</span><span class="sc">;</span>
|
|
397 384 <span class="s">}</span>
|
|
398 385 <span class="s">}</span>
|
|
399 386
|
|
400 387 <span class="c"># Process specified start column value...</span>
|
|
401 <a name="ProcessStartColInfo-"></a> 388 <span class="k">sub </span><span class="m">ProcessStartColInfo</span> <span class="s">{</span>
|
|
402 389 <span class="k">my</span><span class="s">(</span><span class="i">$Index</span><span class="cm">,</span> <span class="i">$ColIndex</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$StartColNum</span><span class="cm">,</span> <span class="i">$Part1StartColNum</span><span class="cm">,</span> <span class="i">$Part1EndColNum</span><span class="cm">,</span> <span class="i">$Part2StartColNum</span><span class="cm">,</span> <span class="i">$Part2EndColNum</span><span class="cm">,</span> <span class="i">$BeforeStartColNum</span><span class="cm">,</span> <span class="i">$AfterStartColNum</span><span class="cm">,</span> <span class="i">$FirstColNum</span><span class="cm">,</span> <span class="i">$LastColNum</span><span class="cm">,</span> <span class="i">$FirstIndex</span><span class="cm">,</span> <span class="i">$LastIndex</span><span class="s">)</span><span class="sc">;</span>
|
|
403 390
|
|
404 391 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part1ColNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
405 392 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part2ColNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
406 393
|
|
407 394 <span class="i">$StartColNum</span> = <span class="q">"last"</span><span class="sc">;</span>
|
|
408 395 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">StartCol</span>}<span class="s">)</span> <span class="s">{</span>
|
|
409 396 <span class="k">if</span> <span class="s">(</span><span class="k">length</span><span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">StartCol</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
410 397 <span class="i">$StartColNum</span> = <span class="i">$OptionsInfo</span>{<span class="w">StartCol</span>}
|
|
411 398 <span class="s">}</span>
|
|
412 399 <span class="s">}</span>
|
|
413 400
|
|
414 401 <span class="k">if</span> <span class="s">(</span><span class="i">$StartColNum</span> !~ <span class="q">/^last$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
415 402 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} =~ <span class="q">/^collabel$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
416 403 <span class="k">if</span> <span class="s">(</span><span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="n">0</span>]{<span class="i">$StartColNum</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
417 404 <span class="i">$StartColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="n">0</span>]{<span class="i">$StartColNum</span>}<span class="sc">;</span>
|
|
418 405 <span class="s">}</span>
|
|
419 406 <span class="k">else</span> <span class="s">{</span>
|
|
420 407 <span class="k">die</span> <span class="q">"Error: Invalid value $StartColNum specified using \"-s --startcol\" option: column name doesn't exist in $TextFilesList[0] \n"</span><span class="sc">;</span>
|
|
421 408 <span class="s">}</span>
|
|
422 409 <span class="s">}</span>
|
|
423 410 <span class="k">else</span> <span class="s">{</span>
|
|
424 411 <span class="k">if</span> <span class="s">(</span><span class="i">$StartColNum</span> > <span class="n">0</span> && <span class="i">$StartColNum</span> <= <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="n">0</span>]<span class="s">)</span> <span class="s">{</span>
|
|
425 412 <span class="i">$StartColNum</span> -= <span class="n">1</span><span class="sc">;</span>
|
|
426 413 <span class="s">}</span>
|
|
427 414 <span class="k">else</span> <span class="s">{</span>
|
|
428 415 <span class="k">die</span> <span class="q">"Error: Invalid value $StartColNum specified using \"-s --startcol\" option: column number doesn't exist in $TextFilesList[0] \n"</span><span class="sc">;</span>
|
|
429 416 <span class="s">}</span>
|
|
430 417 <span class="s">}</span>
|
|
431 418 <span class="s">}</span>
|
|
432 419 <span class="k">else</span> <span class="s">{</span>
|
|
433 420 <span class="i">$StartColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="n">0</span>] - <span class="n">1</span><span class="sc">;</span>
|
|
434 421 <span class="s">}</span>
|
|
435 422
|
|
436 423 <span class="c"># Make sure StartColNum is present on the list of columns to merge for the first text file...</span>
|
|
437 424 <span class="k">if</span> <span class="s">(</span>!<span class="k">exists</span><span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColToMergeNumToLabelMap</span>}[<span class="n">0</span>]{<span class="i">$StartColNum</span>}<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
438 425 <span class="k">die</span> <span class="q">"Error: Invalid value $StartColNum specified using \"-s --startcol\" option: doesn't exist in the specified lists of columns to merge for $TextFilesList[0] \n"</span><span class="sc">;</span>
|
|
439 426 <span class="s">}</span>
|
|
440 427
|
|
441 428 <span class="c"># Find out the column number before and after StartColNum in first text file...</span>
|
|
442 429 <span class="i">$BeforeStartColNum</span> = <span class="i">$StartColNum</span><span class="sc">;</span>
|
|
443 430 <span class="i">$AfterStartColNum</span> = <span class="i">$StartColNum</span><span class="sc">;</span>
|
|
444 431
|
|
445 432 <span class="i">$FirstIndex</span> = <span class="n">0</span><span class="sc">;</span> <span class="i">$LastIndex</span> = <span class="i">$#</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>]}<span class="sc">;</span>
|
|
446 433
|
|
447 434 <span class="i">$FirstColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$FirstIndex</span>]<span class="sc">;</span>
|
|
448 435 <span class="i">$LastColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$LastIndex</span>]<span class="sc">;</span>
|
|
449 436
|
|
450 437 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$LastIndex</span><span class="s">)</span> <span class="s">{</span>
|
|
451 438 <span class="k">if</span> <span class="s">(</span><span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$Index</span>] == <span class="i">$StartColNum</span><span class="s">)</span> <span class="s">{</span>
|
|
452 439 <span class="i">$BeforeStartColNum</span> = <span class="s">(</span><span class="s">(</span><span class="i">$Index</span> -<span class="n">1</span><span class="s">)</span> >= <span class="i">$FirstIndex</span><span class="s">)</span> ? <span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$Index</span> - <span class="n">1</span>] <span class="co">:</span> <span class="s">(</span><span class="i">$FirstColNum</span> - <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
453 440 <span class="i">$AfterStartColNum</span> = <span class="s">(</span><span class="s">(</span><span class="i">$Index</span> + <span class="n">1</span><span class="s">)</span> <= <span class="i">$LastIndex</span><span class="s">)</span> ? <span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$Index</span> + <span class="n">1</span>] <span class="co">:</span> <span class="s">(</span><span class="i">$LastColNum</span> + <span class="n">1</span><span class="s">)</span><span class="sc">;</span>
|
|
454 441 <span class="s">}</span>
|
|
455 442 <span class="s">}</span>
|
|
456 443
|
|
457 444 <span class="k">if</span> <span class="s">(</span><span class="i">$OptionsInfo</span>{<span class="w">StartColMode</span>} =~ <span class="q">/^after$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
458 445 <span class="i">$Part1StartColNum</span> = <span class="i">$FirstColNum</span><span class="sc">;</span> <span class="i">$Part1EndColNum</span> = <span class="i">$StartColNum</span><span class="sc">;</span>
|
|
459 446 <span class="i">$Part2StartColNum</span> = <span class="i">$AfterStartColNum</span><span class="sc">;</span> <span class="i">$Part2EndColNum</span> = <span class="i">$LastColNum</span><span class="sc">;</span>
|
|
460 447 <span class="s">}</span>
|
|
461 448 <span class="k">else</span> <span class="s">{</span>
|
|
462 449 <span class="i">$Part1StartColNum</span> = <span class="i">$FirstColNum</span><span class="sc">;</span> <span class="i">$Part1EndColNum</span> = <span class="i">$BeforeStartColNum</span><span class="sc">;</span>
|
|
463 450 <span class="i">$Part2StartColNum</span> = <span class="i">$StartColNum</span><span class="sc">;</span> <span class="i">$Part2EndColNum</span> = <span class="i">$LastColNum</span><span class="sc">;</span>
|
|
464 451 <span class="s">}</span>
|
|
465 452
|
|
466 453 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part1ColNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
467 454 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part2ColNums</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
468 455
|
|
469 456 <span class="k">for</span> <span class="i">$ColIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
470 457 <span class="i">$ColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
471 458 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> >= <span class="i">$Part1StartColNum</span> && <span class="i">$ColNum</span> <= <span class="i">$Part1EndColNum</span><span class="s">)</span> <span class="s">{</span>
|
|
472 459 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part1ColNums</span>}}<span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
473 460 <span class="s">}</span>
|
|
474 461 <span class="s">}</span>
|
|
475 462
|
|
476 463 <span class="k">for</span> <span class="i">$ColIndex</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>]}<span class="s">)</span> <span class="s">{</span>
|
|
477 464 <span class="i">$ColNum</span> = <span class="i">$TextFilesInfo</span>{<span class="w">ColToMerge</span>}[<span class="n">0</span>][<span class="i">$ColIndex</span>]<span class="sc">;</span>
|
|
478 465 <span class="k">if</span> <span class="s">(</span><span class="i">$ColNum</span> >= <span class="i">$Part2StartColNum</span> && <span class="i">$ColNum</span> <= <span class="i">$Part2EndColNum</span><span class="s">)</span> <span class="s">{</span>
|
|
479 466 <span class="k">push</span> <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">File1Part2ColNums</span>}}<span class="cm">,</span> <span class="i">$ColNum</span><span class="sc">;</span>
|
|
480 467 <span class="s">}</span>
|
|
481 468 <span class="s">}</span>
|
|
482 469
|
|
483 470 <span class="s">}</span>
|
|
484 471
|
|
485 472 <span class="c"># Retrieve information about input text files...</span>
|
|
486 <a name="RetrieveTextFilesInfo-"></a> 473 <span class="k">sub </span><span class="m">RetrieveTextFilesInfo</span> <span class="s">{</span>
|
|
487 474 <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">$Line</span><span class="cm">,</span> <span class="i">$ColNum</span><span class="cm">,</span> <span class="i">$ColLabel</span><span class="cm">,</span> <span class="i">$FileNotOkayCount</span><span class="cm">,</span> <span class="i">@ColLabels</span><span class="cm">,</span><span class="s">)</span><span class="sc">;</span>
|
|
488 475
|
|
489 476 <span class="i">%TextFilesInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
490 477
|
|
491 478 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
492 479 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
493 480 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
494 481 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
495 482 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
496 483
|
|
497 484 <span class="i">$FileNotOkayCount</span> = <span class="n">0</span><span class="sc">;</span>
|
|
498 485
|
|
499 486 <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>
|
|
500 487 <span class="i">$TextFile</span> = <span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="sc">;</span>
|
|
501 488
|
|
502 489 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
503 490 <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] = <span class="n">0</span><span class="sc">;</span>
|
|
504 491 <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="q">""</span><span class="sc">;</span>
|
|
505 492
|
|
506 493 <span class="i">@</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabels</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
507 494 <span class="i">%</span>{<span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
508 495
|
|
509 496 <span class="k">if</span> <span class="s">(</span>!<span class="s">(</span><span class="k">-e</span> <span class="i">$TextFile</span><span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
510 497 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: It doesn't exist\n"</span><span class="sc">;</span>
|
|
511 498 <span class="i">$FileNotOkayCount</span>++<span class="sc">;</span>
|
|
512 499 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
513 500 <span class="s">}</span>
|
|
514 501 <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>
|
|
515 502 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: It's not a csv or tsv file\n"</span><span class="sc">;</span>
|
|
516 503 <span class="i">$FileNotOkayCount</span>++<span class="sc">;</span>
|
|
517 504 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
518 505 <span class="s">}</span>
|
|
519 506 <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>
|
|
520 507 <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>
|
|
521 508 <span class="i">$InDelim</span> = <span class="q">"\t"</span><span class="sc">;</span>
|
|
522 509 <span class="s">}</span>
|
|
523 510 <span class="k">else</span> <span class="s">{</span>
|
|
524 511 <span class="i">$InDelim</span> = <span class="q">"\,"</span><span class="sc">;</span>
|
|
525 512 <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>
|
|
526 513 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: The value specified, $OptionsInfo{InDelim}, for option \"--indelim\" is not valid for csv files\n"</span><span class="sc">;</span>
|
|
527 514 <span class="i">$FileNotOkayCount</span>++<span class="sc">;</span>
|
|
528 515 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
529 516 <span class="s">}</span>
|
|
530 517 <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>
|
|
531 518 <span class="i">$InDelim</span> = <span class="q">"\;"</span><span class="sc">;</span>
|
|
532 519 <span class="s">}</span>
|
|
533 520 <span class="s">}</span>
|
|
534 521
|
|
535 522 <span class="k">if</span> <span class="s">(</span>!<span class="k">open</span> <span class="w">TEXTFILE</span><span class="cm">,</span> <span class="q">"$TextFile"</span><span class="s">)</span> <span class="s">{</span>
|
|
536 523 <span class="k">warn</span> <span class="q">"Warning: Ignoring file $TextFile: Couldn't open it: $! \n"</span><span class="sc">;</span>
|
|
537 524 <span class="i">$FileNotOkayCount</span>++<span class="sc">;</span>
|
|
538 525 <span class="k">next</span> <span class="j">FILELIST</span><span class="sc">;</span>
|
|
539 526 <span class="s">}</span>
|
|
540 527
|
|
541 528 <span class="i">$Line</span> = <span class="i">GetTextLine</span><span class="s">(</span>\<span class="i">*TEXTFILE</span><span class="s">)</span><span class="sc">;</span>
|
|
542 529 <span class="i">@ColLabels</span> = <span class="i">quotewords</span><span class="s">(</span><span class="i">$InDelim</span><span class="cm">,</span> <span class="n">0</span><span class="cm">,</span> <span class="i">$Line</span><span class="s">)</span><span class="sc">;</span>
|
|
543 530 <span class="k">close</span> <span class="w">TEXTFILE</span><span class="sc">;</span>
|
|
544 531
|
|
545 532 <span class="i">$TextFilesInfo</span>{<span class="w">FileOkay</span>}[<span class="i">$Index</span>] = <span class="n">1</span><span class="sc">;</span>
|
|
546 533 <span class="i">$TextFilesInfo</span>{<span class="w">InDelim</span>}[<span class="i">$Index</span>] = <span class="i">$InDelim</span><span class="sc">;</span>
|
|
547 534
|
|
548 535 <span class="i">$TextFilesInfo</span>{<span class="w">ColCount</span>}[<span class="i">$Index</span>] = <span class="i">@ColLabels</span><span class="sc">;</span>
|
|
549 536 <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>
|
|
550 537 <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>
|
|
551 538 <span class="i">$ColLabel</span> = <span class="i">$ColLabels</span>[<span class="i">$ColNum</span>]<span class="sc">;</span>
|
|
552 539 <span class="i">$TextFilesInfo</span>{<span class="w">ColLabelToNumMap</span>}[<span class="i">$Index</span>]{<span class="i">$ColLabel</span>} = <span class="i">$ColNum</span><span class="sc">;</span>
|
|
553 540 <span class="s">}</span>
|
|
554 541 <span class="s">}</span>
|
|
555 542 <span class="c"># Make sure all specified files are valid for merging to work properly...</span>
|
|
556 543 <span class="k">if</span> <span class="s">(</span><span class="i">$FileNotOkayCount</span><span class="s">)</span> <span class="s">{</span>
|
|
557 544 <span class="k">die</span> <span class="q">"Error: Problems with input text file(s)...\n"</span><span class="sc">;</span>
|
|
558 545 <span class="s">}</span>
|
|
559 546 <span class="s">}</span>
|
|
560 547
|
|
561 548 <span class="c"># Process option values...</span>
|
|
562 <a name="ProcessOptions-"></a> 549 <span class="k">sub </span><span class="m">ProcessOptions</span> <span class="s">{</span>
|
|
563 550 <span class="k">my</span><span class="s">(</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">$NewTextFile</span><span class="cm">,</span> <span class="i">@ColValues</span><span class="cm">,</span> <span class="i">@KeyValues</span><span class="s">)</span><span class="sc">;</span>
|
|
564 551
|
|
565 552 <span class="i">%OptionsInfo</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
566 553
|
|
567 554 <span class="i">$OptionsInfo</span>{<span class="w">Mode</span>} = <span class="i">$Options</span>{<span class="w">mode</span>}<span class="sc">;</span>
|
|
568 555
|
|
569 556 <span class="i">$OptionsInfo</span>{<span class="w">Columns</span>} = <span class="i">$Options</span>{<span class="w">columns</span>}<span class="sc">;</span>
|
|
570 557 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">ColValues</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
571 558
|
|
572 559 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">columns</span>}<span class="s">)</span> <span class="s">{</span>
|
|
573 560 <span class="i">@ColValues</span> = <span class="k">split</span> <span class="q">";"</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">columns</span>}<span class="sc">;</span>
|
|
574 561 <span class="k">if</span> <span class="s">(</span><span class="i">@ColValues</span> != <span class="i">@TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
575 562 <span class="k">die</span> <span class="q">"Error: Invalid number of values specified by \"-c --columns\" option: it must be equal to number of input text files.\n"</span><span class="sc">;</span>
|
|
576 563 <span class="s">}</span>
|
|
577 564 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#ColValues</span><span class="s">)</span> <span class="s">{</span>
|
|
578 565 <span class="k">if</span> <span class="s">(</span>!<span class="k">length</span><span class="s">(</span><span class="i">$ColValues</span>[<span class="i">$Index</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
579 566 <span class="k">die</span> <span class="q">"Error: Invalid value specified by \"-c --columns\" option: empty values are not allowed.\n"</span><span class="sc">;</span>
|
|
580 567 <span class="s">}</span>
|
|
581 568 <span class="s">}</span>
|
|
582 569 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">ColValues</span>}} = <span class="i">@ColValues</span><span class="sc">;</span>
|
|
583 570 <span class="s">}</span>
|
|
584 571
|
|
585 572 <span class="i">$OptionsInfo</span>{<span class="w">Keys</span>} = <span class="i">$Options</span>{<span class="w">keys</span>}<span class="sc">;</span>
|
|
586 573 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">KeyValues</span>}} = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
587 574
|
|
588 575 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">keys</span>}<span class="s">)</span> <span class="s">{</span>
|
|
589 576 <span class="i">@KeyValues</span> = <span class="k">split</span> <span class="q">";"</span><span class="cm">,</span> <span class="i">$Options</span>{<span class="w">keys</span>}<span class="sc">;</span>
|
|
590 577 <span class="k">if</span> <span class="s">(</span><span class="i">@KeyValues</span> != <span class="i">@TextFilesList</span><span class="s">)</span> <span class="s">{</span>
|
|
591 578 <span class="k">die</span> <span class="q">"Error: Invalid number of values specified by \"-k --keys\" option: it must be equal to number of input text files.\n"</span><span class="sc">;</span>
|
|
592 579 <span class="s">}</span>
|
|
593 580 <span class="k">for</span> <span class="i">$Index</span> <span class="s">(</span><span class="n">0</span> .. <span class="i">$#KeyValues</span><span class="s">)</span> <span class="s">{</span>
|
|
594 581 <span class="k">if</span> <span class="s">(</span>!<span class="k">length</span><span class="s">(</span><span class="i">$KeyValues</span>[<span class="i">$Index</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
595 582 <span class="k">die</span> <span class="q">"Error: Invalid value specified by \"-k --keys\" option: empty values are not allowed.\n"</span><span class="sc">;</span>
|
|
596 583 <span class="s">}</span>
|
|
597 584 <span class="s">}</span>
|
|
598 585 <span class="i">@</span>{<span class="i">$OptionsInfo</span>{<span class="w">KeyValues</span>}} = <span class="i">@KeyValues</span><span class="sc">;</span>
|
|
599 586 <span class="s">}</span>
|
|
600 587
|
|
601 588 <span class="i">$OptionsInfo</span>{<span class="w">InDelim</span>} = <span class="i">$Options</span>{<span class="w">indelim</span>}<span class="sc">;</span>
|
|
602 589
|
|
603 590 <span class="i">$OptionsInfo</span>{<span class="w">StartCol</span>} = <span class="i">$Options</span>{<span class="w">startcol</span>} ? <span class="i">$Options</span>{<span class="w">startcol</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
604 591 <span class="i">$OptionsInfo</span>{<span class="w">StartColMode</span>} = <span class="i">$Options</span>{<span class="w">startcolmode</span>}<span class="sc">;</span>
|
|
605 592
|
|
606 593 <span class="i">$OptionsInfo</span>{<span class="w">OutFileRoot</span>} = <span class="i">$Options</span>{<span class="w">root</span>} ? <span class="i">$Options</span>{<span class="w">root</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
607 594 <span class="i">$OptionsInfo</span>{<span class="w">Overwrite</span>} = <span class="i">$Options</span>{<span class="w">overwrite</span>} ? <span class="i">$Options</span>{<span class="w">overwrite</span>} <span class="co">:</span> <span class="k">undef</span><span class="sc">;</span>
|
|
608 595
|
|
609 596 <span class="i">$OptionsInfo</span>{<span class="w">OutDelim</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span> <span class="s">)</span> ? <span class="q">"\t"</span> <span class="co">:</span> <span class="s">(</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^semicolon$/i</span><span class="s">)</span> ? <span class="q">"\;"</span> <span class="co">:</span> <span class="q">"\,"</span><span class="s">)</span><span class="sc">;</span>
|
|
610 597 <span class="i">$OptionsInfo</span>{<span class="w">OutQuote</span>} = <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} =~ <span class="q">/^yes$/i</span><span class="s">)</span> ? <span class="n">1</span> <span class="co">:</span> <span class="n">0</span><span class="sc">;</span>
|
|
611 598
|
|
612 599 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">root</span>}<span class="s">)</span> <span class="s">{</span>
|
|
613 600 <span class="i">$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>
|
|
614 601 <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">$FileExt</span><span class="s">)</span> = <span class="i">ParseFileName</span><span class="s">(</span><span class="i">$Options</span>{<span class="w">root</span>}<span class="s">)</span><span class="sc">;</span>
|
|
615 602 <span class="k">if</span> <span class="s">(</span><span class="i">$FileName</span> && <span class="i">$FileExt</span><span class="s">)</span> <span class="s">{</span>
|
|
616 603 <span class="i">$NewTextFile</span> = <span class="i">$FileName</span><span class="sc">;</span>
|
|
617 604 <span class="s">}</span> <span class="k">else</span> <span class="s">{</span>
|
|
618 605 <span class="i">$NewTextFile</span> = <span class="i">$Options</span>{<span class="w">root</span>}<span class="sc">;</span>
|
|
619 606 <span class="s">}</span>
|
|
620 607 <span class="s">}</span> <span class="k">else</span> <span class="s">{</span>
|
|
621 608 <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">$FileExt</span> = <span class="q">""</span><span class="sc">;</span>
|
|
622 609 <span class="s">(</span><span class="i">$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">$TextFilesList</span>[<span class="n">0</span>]<span class="s">)</span><span class="sc">;</span>
|
|
623 610 <span class="i">$NewTextFile</span> = <span class="i">$FileName</span> . <span class="q">"1To"</span> . <span class="i">@TextFilesList</span> . <span class="q">"Merged"</span><span class="sc">;</span>
|
|
624 611 <span class="s">}</span>
|
|
625 612 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} =~ <span class="q">/^tab$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
626 613 <span class="i">$NewTextFile</span> .= <span class="q">".tsv"</span><span class="sc">;</span>
|
|
627 614 <span class="s">}</span> <span class="k">else</span> <span class="s">{</span>
|
|
628 615 <span class="i">$NewTextFile</span> .= <span class="q">".csv"</span><span class="sc">;</span>
|
|
629 616 <span class="s">}</span>
|
|
630 617 <span class="k">if</span> <span class="s">(</span>!<span class="i">$Options</span>{<span class="w">overwrite</span>}<span class="s">)</span> <span class="s">{</span>
|
|
631 618 <span class="k">if</span> <span class="s">(</span><span class="k">-e</span> <span class="i">$NewTextFile</span><span class="s">)</span> <span class="s">{</span>
|
|
632 619 <span class="k">die</span> <span class="q">"Error: The file $NewTextFile already exists.\n"</span><span class="sc">;</span>
|
|
633 620 <span class="s">}</span>
|
|
634 621 <span class="s">}</span>
|
|
635 622 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">root</span>}<span class="s">)</span> <span class="s">{</span>
|
|
636 623 <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>
|
|
637 624 <span class="k">if</span> <span class="s">(</span><span class="k">lc</span><span class="s">(</span><span class="i">$NewTextFile</span><span class="s">)</span> <span class="k">eq</span> <span class="k">lc</span><span class="s">(</span><span class="i">$TextFilesList</span>[<span class="i">$Index</span>]<span class="s">)</span><span class="s">)</span> <span class="s">{</span>
|
|
638 625 <span class="k">die</span> <span class="q">"Error: Output filename, $NewTextFile, is similar to a input file name.\nSpecify a different name using \"-r --root\" option or use default name.\n"</span><span class="sc">;</span>
|
|
639 626 <span class="s">}</span>
|
|
640 627 <span class="s">}</span>
|
|
641 628 <span class="s">}</span>
|
|
642 629
|
|
643 630 <span class="i">$OptionsInfo</span>{<span class="w">NewTextFile</span>} = <span class="i">$NewTextFile</span><span class="sc">;</span>
|
|
644 631 <span class="s">}</span>
|
|
645 632
|
|
646 633 <span class="c"># Setup script usage and retrieve command line arguments specified using various options...</span>
|
|
647 <a name="SetupScriptUsage-"></a> 634 <span class="k">sub </span><span class="m">SetupScriptUsage</span> <span class="s">{</span>
|
|
648 635
|
|
649 636 <span class="c"># Retrieve all the options...</span>
|
|
650 637 <span class="i">%Options</span> = <span class="s">(</span><span class="s">)</span><span class="sc">;</span>
|
|
651 638
|
|
652 639 <span class="i">$Options</span>{<span class="w">mode</span>} = <span class="q">"colnum"</span><span class="sc">;</span>
|
|
653 640 <span class="i">$Options</span>{<span class="w">indelim</span>} = <span class="q">"comma"</span><span class="sc">;</span>
|
|
654 641 <span class="i">$Options</span>{<span class="w">outdelim</span>} = <span class="q">"comma"</span><span class="sc">;</span>
|
|
655 642 <span class="i">$Options</span>{<span class="w">quote</span>} = <span class="q">"yes"</span><span class="sc">;</span>
|
|
656 643 <span class="i">$Options</span>{<span class="w">startcolmode</span>} = <span class="q">"after"</span><span class="sc">;</span>
|
|
657 644
|
|
658 645 <span class="k">if</span> <span class="s">(</span>!<span class="i">GetOptions</span><span class="s">(</span>\<span class="i">%Options</span><span class="cm">,</span> <span class="q">"help|h"</span><span class="cm">,</span> <span class="q">"indelim=s"</span><span class="cm">,</span> <span class="q">"columns|c=s"</span><span class="cm">,</span> <span class="q">"keys|k=s"</span><span class="cm">,</span> <span class="q">"mode|m=s"</span><span class="cm">,</span> <span class="q">"outdelim=s"</span><span class="cm">,</span> <span class="q">"overwrite|o"</span><span class="cm">,</span> <span class="q">"quote|q=s"</span><span class="cm">,</span> <span class="q">"root|r=s"</span><span class="cm">,</span> <span class="q">"startcol|s=s"</span><span class="cm">,</span> <span class="q">"startcolmode=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>
|
|
659 646 <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 $ScriptName -h\" command and try again...\n"</span><span class="sc">;</span>
|
|
660 647 <span class="s">}</span>
|
|
661 648 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">workingdir</span>}<span class="s">)</span> <span class="s">{</span>
|
|
662 649 <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>
|
|
663 650 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{workingdir}, for option \"-w --workingdir\" is not a directory name.\n"</span><span class="sc">;</span>
|
|
664 651 <span class="s">}</span>
|
|
665 652 <span class="k">chdir</span> <span class="i">$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}: $! \n"</span><span class="sc">;</span>
|
|
666 653 <span class="s">}</span>
|
|
667 654 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">mode</span>} !~ <span class="q">/^(colnum|collabel)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
668 655 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{mode}, for option \"-m --mode\" is not valid. Allowed values: colnum, or collabel\n"</span><span class="sc">;</span>
|
|
669 656 <span class="s">}</span>
|
|
670 657 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">indelim</span>} !~ <span class="q">/^(comma|semicolon)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
671 658 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{indelim}, for option \"--indelim\" is not valid. Allowed values: comma or semicolon\n"</span><span class="sc">;</span>
|
|
672 659 <span class="s">}</span>
|
|
673 660 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">outdelim</span>} !~ <span class="q">/^(comma|semicolon|tab)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
674 661 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{outdelim}, for option \"--outdelim\" is not valid. Allowed values: comma, tab, or semicolon\n"</span><span class="sc">;</span>
|
|
675 662 <span class="s">}</span>
|
|
676 663 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">quote</span>} !~ <span class="q">/^(yes|no)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
677 664 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"-q --quote\" is not valid. Allowed values: yes or no\n"</span><span class="sc">;</span>
|
|
678 665 <span class="s">}</span>
|
|
679 666 <span class="k">if</span> <span class="s">(</span><span class="i">$Options</span>{<span class="w">startcolmode</span>} !~ <span class="q">/^(before|after)$/i</span><span class="s">)</span> <span class="s">{</span>
|
|
680 667 <span class="k">die</span> <span class="q">"Error: The value specified, $Options{quote}, for option \"--startcolmode\" is not valid. Allowed values: before or after\n"</span><span class="sc">;</span>
|
|
681 668 <span class="s">}</span>
|
|
682 669 <span class="s">}</span>
|
|
683 670
|
|
684 <a name="EOF-"></a></pre>
|
|
685 <p> </p>
|
|
686 <br />
|
|
687 <center>
|
|
688 <img src="../../../images/h2o2.png">
|
|
689 </center>
|
|
690 </body>
|
|
691 </html>
|