1 |
joko |
1.1 |
## --------------------------------------------------------------------------- |
2 |
root |
1.2 |
## $Id: String.pm,v 1.1 2003/01/19 03:27:36 joko Exp $ |
3 |
joko |
1.1 |
## --------------------------------------------------------------------------- |
4 |
root |
1.2 |
## $Log: String.pm,v $ |
5 |
|
|
## Revision 1.1 2003/01/19 03:27:36 joko |
6 |
|
|
## + initial check-in |
7 |
|
|
## |
8 |
joko |
1.1 |
## --------------------------------------------------------------------------- |
9 |
|
|
|
10 |
|
|
|
11 |
|
|
package Data::Transform::String; |
12 |
|
|
|
13 |
|
|
use strict; |
14 |
|
|
use warnings; |
15 |
|
|
|
16 |
|
|
require Exporter; |
17 |
|
|
our @ISA = qw( Exporter ); |
18 |
|
|
our @EXPORT_OK = qw( |
19 |
root |
1.2 |
stripHtml stripSpaces stripNewLines toReal trim getLastPart |
20 |
joko |
1.1 |
); |
21 |
|
|
|
22 |
|
|
sub stripHtml { |
23 |
|
|
my $html = shift; |
24 |
|
|
my $result = ''; |
25 |
|
|
#$html =~ s/<br>(.*)/ - ($1)/i; |
26 |
|
|
my $p = HTML::PullParser->new( |
27 |
|
|
doc => \$html, |
28 |
|
|
text => 'text', |
29 |
|
|
unbroken_text => 1, |
30 |
|
|
); |
31 |
|
|
while (my $token = $p->get_token()) { |
32 |
|
|
my $text = join('', @{$token}); |
33 |
|
|
$result .= $text; |
34 |
|
|
} |
35 |
|
|
#$result =~ s/ //g; |
36 |
|
|
return $result; |
37 |
|
|
} |
38 |
|
|
|
39 |
|
|
sub stripSpaces { |
40 |
|
|
my $text = shift; |
41 |
|
|
#print "text: $text", "\n"; |
42 |
|
|
#print "ord: ", ord(substr($text, 0, 1)), "\n"; |
43 |
|
|
$text =~ s/^\s*//g; |
44 |
|
|
$text =~ s/\s*$//g; |
45 |
|
|
return $text; |
46 |
|
|
} |
47 |
|
|
|
48 |
|
|
sub trim { |
49 |
|
|
my $string = shift; |
50 |
|
|
return stripSpaces($string); |
51 |
|
|
} |
52 |
|
|
|
53 |
|
|
sub stripNewLines { |
54 |
|
|
my $text = shift; |
55 |
|
|
#print "text: $text", "\n"; |
56 |
|
|
#print "ord: ", ord(substr($text, 0, 1)), "\n"; |
57 |
|
|
$text =~ s/\n//g; |
58 |
|
|
#$text =~ s/\s*$//g; |
59 |
|
|
return $text; |
60 |
|
|
} |
61 |
|
|
|
62 |
|
|
sub toReal { |
63 |
|
|
my $string = shift; |
64 |
|
|
$string =~ m/(\d+\.*\d+)/; |
65 |
|
|
my $real = $1; |
66 |
|
|
return $real; |
67 |
root |
1.2 |
} |
68 |
|
|
|
69 |
|
|
sub getLastPart { |
70 |
|
|
my $which = shift; |
71 |
|
|
my $seperator = shift; |
72 |
|
|
$seperator ||= '\/'; |
73 |
|
|
$which =~ m/^.*$seperator(.+?)$/; |
74 |
|
|
return $1; |
75 |
joko |
1.1 |
} |
76 |
|
|
|
77 |
|
|
1; |