| 3 |
# $Id$ |
# $Id$ |
| 4 |
# |
# |
| 5 |
# $Log$ |
# $Log$ |
| 6 |
|
# Revision 1.5 2002/10/27 18:34:28 joko |
| 7 |
|
# + sub now |
| 8 |
|
# |
| 9 |
|
# Revision 1.4 2002/08/16 19:06:39 cvsjoko |
| 10 |
|
# + sub getDirList |
| 11 |
|
# |
| 12 |
|
# Revision 1.3 2002/07/19 18:13:50 cvsjoko |
| 13 |
|
# no message |
| 14 |
|
# |
| 15 |
|
# Revision 1.2 2002/06/27 02:14:22 cvsjoko |
| 16 |
|
# + stripHtml stripSpaces stripNewLines toReal |
| 17 |
|
# |
| 18 |
# Revision 1.1 2002/06/24 14:49:59 cvsjoko |
# Revision 1.1 2002/06/24 14:49:59 cvsjoko |
| 19 |
# + new |
# + new |
| 20 |
# |
# |
| 26 |
require Exporter; |
require Exporter; |
| 27 |
@ISA = qw( Exporter ); |
@ISA = qw( Exporter ); |
| 28 |
@EXPORT = qw( |
@EXPORT = qw( |
| 29 |
Dumper |
Dumper |
| 30 |
md5 md5_hex md5_base64 |
md5 md5_hex md5_base64 |
| 31 |
ParseDate UnixDate |
ParseDate UnixDate |
| 32 |
|
strftime |
| 33 |
|
stripHtml stripSpaces stripNewLines toReal trim |
| 34 |
|
croak |
| 35 |
|
array_getDifference |
| 36 |
|
getDirList |
| 37 |
|
now |
| 38 |
); |
); |
| 39 |
|
|
| 40 |
use strict; |
use strict; |
| 46 |
$main::TZ = 'GMT'; |
$main::TZ = 'GMT'; |
| 47 |
use Date::Manip; |
use Date::Manip; |
| 48 |
|
|
| 49 |
|
require LWP::UserAgent; |
| 50 |
|
use HTML::PullParser; |
| 51 |
|
|
| 52 |
|
# $now_string = strftime "%a %b %e %H:%M:%S %Y", localtime; |
| 53 |
|
# see "perldoc -f localtime" |
| 54 |
|
use POSIX qw(strftime); |
| 55 |
|
|
| 56 |
|
use Carp; |
| 57 |
|
|
| 58 |
|
use DirHandle; |
| 59 |
|
|
| 60 |
|
|
| 61 |
|
######################################## |
| 62 |
|
|
| 63 |
|
sub stripSpaces { |
| 64 |
|
my $text = shift; |
| 65 |
|
#print "text: $text", "\n"; |
| 66 |
|
#print "ord: ", ord(substr($text, 0, 1)), "\n"; |
| 67 |
|
$text =~ s/^\s*//g; |
| 68 |
|
$text =~ s/\s*$//g; |
| 69 |
|
return $text; |
| 70 |
|
} |
| 71 |
|
|
| 72 |
|
sub trim { |
| 73 |
|
my $string = shift; |
| 74 |
|
return stripSpaces($string); |
| 75 |
|
} |
| 76 |
|
|
| 77 |
|
sub stripNewLines { |
| 78 |
|
my $text = shift; |
| 79 |
|
#print "text: $text", "\n"; |
| 80 |
|
#print "ord: ", ord(substr($text, 0, 1)), "\n"; |
| 81 |
|
$text =~ s/\n//g; |
| 82 |
|
#$text =~ s/\s*$//g; |
| 83 |
|
return $text; |
| 84 |
|
} |
| 85 |
|
|
| 86 |
|
sub toReal { |
| 87 |
|
my $string = shift; |
| 88 |
|
$string =~ m/(\d+\.*\d+)/; |
| 89 |
|
my $real = $1; |
| 90 |
|
return $real; |
| 91 |
|
} |
| 92 |
|
|
| 93 |
|
sub stripHtml { |
| 94 |
|
my $html = shift; |
| 95 |
|
my $result = ''; |
| 96 |
|
#$html =~ s/<br>(.*)/ - ($1)/i; |
| 97 |
|
my $p = HTML::PullParser->new( |
| 98 |
|
doc => \$html, |
| 99 |
|
text => 'text', |
| 100 |
|
unbroken_text => 1, |
| 101 |
|
); |
| 102 |
|
while (my $token = $p->get_token()) { |
| 103 |
|
my $text = join('', @{$token}); |
| 104 |
|
$result .= $text; |
| 105 |
|
} |
| 106 |
|
#$result =~ s/ //g; |
| 107 |
|
return $result; |
| 108 |
|
} |
| 109 |
|
|
| 110 |
|
sub array_getRelations { |
| 111 |
|
my $a_ref = shift; |
| 112 |
|
my $b_ref = shift; |
| 113 |
|
my @a = @{$a_ref}; |
| 114 |
|
my @b = @{$b_ref}; |
| 115 |
|
|
| 116 |
|
my @isect = my @diff = my @union = (); |
| 117 |
|
my $e; |
| 118 |
|
my %count; |
| 119 |
|
|
| 120 |
|
foreach $e (@a, @b) { $count{$e}++ } |
| 121 |
|
|
| 122 |
|
foreach $e (keys %count) { |
| 123 |
|
push(@union, $e); |
| 124 |
|
push @{ $count{$e} == 2 ? \@isect : \@diff }, $e; |
| 125 |
|
} |
| 126 |
|
|
| 127 |
|
my $result = { |
| 128 |
|
union => \@union, |
| 129 |
|
isect => \@isect, |
| 130 |
|
diff => \@diff, |
| 131 |
|
}; |
| 132 |
|
|
| 133 |
|
} |
| 134 |
|
|
| 135 |
|
sub array_getDifference { |
| 136 |
|
my $res = array_getRelations(shift, shift); |
| 137 |
|
return $res->{diff}; |
| 138 |
|
} |
| 139 |
|
|
| 140 |
|
|
| 141 |
|
# ============================================= |
| 142 |
|
# "global" vars used in directory-recursive-parsing |
| 143 |
|
my $dirlist_buf; |
| 144 |
|
my @dirlist_path; |
| 145 |
|
my $dirlist_base; |
| 146 |
|
|
| 147 |
|
sub entry_callback { |
| 148 |
|
|
| 149 |
|
my $entry = shift; |
| 150 |
|
|
| 151 |
|
# CHECKS |
| 152 |
|
# dont't use this: |
| 153 |
|
if ($entry eq '.' || $entry eq '..') { return; } |
| 154 |
|
|
| 155 |
|
# PREPARE |
| 156 |
|
# prepare path to current entry |
| 157 |
|
my $cur_entry = join('/', @dirlist_path, $entry); |
| 158 |
|
# prepare path to current entry (absolute) |
| 159 |
|
my $cur_entry_abs = join('/', $dirlist_base, @dirlist_path, $entry); |
| 160 |
|
|
| 161 |
|
# ENTRY |
| 162 |
|
# add current entry to buffer |
| 163 |
|
$dirlist_buf .= $cur_entry . "\n"; |
| 164 |
|
|
| 165 |
|
# (SUB-)DIRECTORY |
| 166 |
|
# check if current entry is a (sub-)directory ... |
| 167 |
|
if (-d $cur_entry_abs) { |
| 168 |
|
push @dirlist_path, $cur_entry; |
| 169 |
|
# ... and parse this (recursion here!!!) |
| 170 |
|
iterate_path($cur_entry_abs); |
| 171 |
|
pop @dirlist_path; |
| 172 |
|
} |
| 173 |
|
} |
| 174 |
|
|
| 175 |
|
sub iterate_path { |
| 176 |
|
|
| 177 |
|
my $path = shift; |
| 178 |
|
|
| 179 |
|
# create new "DirHandle"-object |
| 180 |
|
my $d = new DirHandle $path; |
| 181 |
|
if (defined $d) { |
| 182 |
|
|
| 183 |
|
# iterate through all entries in $path ($d->read) and call out entry-handler on each entry |
| 184 |
|
while (defined(my $line = $d->read)) { |
| 185 |
|
entry_callback($line); |
| 186 |
|
} |
| 187 |
|
|
| 188 |
|
undef $d; |
| 189 |
|
} |
| 190 |
|
} |
| 191 |
|
|
| 192 |
|
sub getDirList { |
| 193 |
|
|
| 194 |
|
$dirlist_base = shift; |
| 195 |
|
|
| 196 |
|
# reset vars |
| 197 |
|
$dirlist_buf = ''; |
| 198 |
|
@dirlist_path = (); |
| 199 |
|
|
| 200 |
|
# start parsing file-structure |
| 201 |
|
iterate_path($dirlist_base); |
| 202 |
|
|
| 203 |
|
# return complete list of directory-content including files and subdirs |
| 204 |
|
# entries are newline (\n) - seperated |
| 205 |
|
return $dirlist_buf; |
| 206 |
|
|
| 207 |
|
} |
| 208 |
|
# ============================================= |
| 209 |
|
|
| 210 |
|
|
| 211 |
|
sub now { |
| 212 |
|
return strftime("%Y-%m-%d %H:%M:%S", localtime); |
| 213 |
|
} |
| 214 |
|
|
| 215 |
1; |
1; |