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