annotate tools/filters/condense_characters.pl @ 0:9071e359b9a3

Uploaded
author xuebing
date Fri, 09 Mar 2012 19:37:19 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
1 #! /usr/bin/perl -w
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
2
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
3 use strict;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
4 use warnings;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
5
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
6 # condenses all consecutive characters of one type
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
7 # convert_characters.pl [input] [character] [output]
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
8
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
9 die "Check arguments" unless @ARGV == 3;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
10
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
11 my $inputfile = $ARGV[0];
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
12 my $character = $ARGV[1];
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
13 my $outputfile = $ARGV[2];
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
14
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
15
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
16 my $convert_from;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
17 my $convert_to;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
18
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
19
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
20 if ($character eq "s")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
21 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
22 $convert_from = '\s';
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
23 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
24 elsif ($character eq "T")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
25 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
26 $convert_from = '\t';
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
27 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
28 elsif ($character eq "Sp")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
29 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
30 $convert_from = " ";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
31 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
32 elsif ($character eq "Dt")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
33 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
34 $convert_from = '\.';
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
35 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
36 elsif ($character eq "C")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
37 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
38 $convert_from = ",";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
39 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
40 elsif ($character eq "D")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
41 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
42 $convert_from = "-";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
43 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
44 elsif ($character eq "U")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
45 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
46 $convert_from = "_";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
47 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
48 elsif ($character eq "P")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
49 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
50 $convert_from = '\|';
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
51 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
52 else
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
53 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
54 die "Invalid value specified for convert from\n";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
55 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
56
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
57
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
58 if ($character eq "T")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
59 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
60 $convert_to = "\t";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
61 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
62 elsif ($character eq "Sp")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
63 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
64 $convert_to = " ";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
65 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
66 elsif ($character eq "Dt")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
67 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
68 $convert_to = "\.";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
69 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
70 elsif ($character eq "C")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
71 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
72 $convert_to = ",";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
73 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
74 elsif ($character eq "D")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
75 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
76 $convert_to = "-";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
77 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
78 elsif ($character eq "U")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
79 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
80 $convert_to = "_";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
81 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
82 elsif ($character eq "P")
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
83 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
84 $convert_to = "|";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
85 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
86 else
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
87 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
88 die "Invalid value specified for Convert to\n";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
89 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
90
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
91 my $fhIn;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
92 open ($fhIn, "< $inputfile") or die "Cannot open source file";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
93
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
94 my $fhOut;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
95 open ($fhOut, "> $outputfile");
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
96
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
97 while (<$fhIn>)
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
98 {
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
99 my $thisLine = $_;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
100 chomp $thisLine;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
101 $thisLine =~ s/${convert_from}+/$convert_to/g;
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
102 print $fhOut $thisLine,"\n";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
103 }
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
104 close ($fhIn) or die "Cannot close source file";
9071e359b9a3 Uploaded
xuebing
parents:
diff changeset
105 close ($fhOut) or die "Cannot close output file";