perl/libs/libp.pm

#################################
#
#  $Id: libp.pm,v 1.5 2002/10/27 18:34:28 joko Exp $
#
#  $Log: libp.pm,v $
#  Revision 1.5  2002/10/27 18:34:28  joko
#  + sub now
#
#  Revision 1.4  2002/08/16 19:06:39  cvsjoko
#  + sub getDirList
#
#  Revision 1.3  2002/07/19 18:13:50  cvsjoko
#  no message
#
#  Revision 1.2  2002/06/27 02:14:22  cvsjoko
#  + stripHtml stripSpaces stripNewLines toReal
#
#  Revision 1.1  2002/06/24 14:49:59  cvsjoko
#  + new
#
#
#################################

package libp;

require Exporter;
@ISA = qw( Exporter );
@EXPORT = qw( 
    Dumper
    md5 md5_hex md5_base64
    ParseDate UnixDate
    strftime
    stripHtml stripSpaces stripNewLines toReal trim
    croak
    array_getDifference
    getDirList
    now
    deep_copy
);

use strict;
use warnings;

use Data::Dumper;
use Digest::MD5 qw(md5 md5_hex md5_base64);

$main::TZ = 'GMT';
use Date::Manip;

require LWP::UserAgent;
use HTML::PullParser;

# $now_string = strftime "%a %b %e %H:%M:%S %Y", localtime;
# see "perldoc -f localtime"
use POSIX qw(strftime);

use Carp;

use DirHandle;


########################################

sub stripSpaces {
  my $text = shift;
  #print "text: $text", "\n";
  #print "ord: ", ord(substr($text, 0, 1)), "\n";
  $text =~ s/^\s*//g;
  $text =~ s/\s*$//g;
  return $text;
}

sub trim {
  my $string = shift;
  return stripSpaces($string);
}

sub stripNewLines {
  my $text = shift;
  #print "text: $text", "\n";
  #print "ord: ", ord(substr($text, 0, 1)), "\n";
  $text =~ s/\n//g;
  #$text =~ s/\s*$//g;
  return $text;
}

sub toReal {
  my $string = shift;
  $string =~ m/(\d+\.*\d+)/;
  my $real = $1;
  return $real;
}

sub stripHtml {
  my $html = shift;
  my $result = '';
  #$html =~ s/<br>(.*)/ - ($1)/i;
  my $p = HTML::PullParser->new(
    doc => \$html,
    text => 'text',
    unbroken_text => 1,
  );
  while (my $token = $p->get_token()) {
    my $text = join('', @{$token});
    $result .= $text;
  }
  #$result =~ s/&nbsp;//g;
  return $result;
}

sub array_getRelations {
  my $a_ref = shift;
  my $b_ref = shift;
  my @a = @{$a_ref};
  my @b = @{$b_ref};

  my @isect = my @diff = my @union = ();
  my $e;
  my %count;
  
  foreach $e (@a, @b) { $count{$e}++ }

  foreach $e (keys %count) {
      push(@union, $e);
      push @{ $count{$e} == 2 ? \@isect : \@diff }, $e;
  }
  
  my $result = {
    union => \@union,
    isect => \@isect,
    diff => \@diff,
  };

}

sub array_getDifference {
  my $res = array_getRelations(shift, shift);
  return $res->{diff};
}


# =============================================
# "global" vars used in directory-recursive-parsing
my $dirlist_buf;
my @dirlist_path;
my $dirlist_base;

sub entry_callback {

  my $entry = shift;

  # CHECKS
  # dont't use this:
  if ($entry eq '.' || $entry eq '..') { return; }

  # PREPARE
  # prepare path to current entry
  my $cur_entry = join('/', @dirlist_path, $entry);
  # prepare path to current entry (absolute)
  my $cur_entry_abs = join('/', $dirlist_base, @dirlist_path, $entry);

  # ENTRY
  # add current entry to buffer
  $dirlist_buf .= $cur_entry . "\n";

  # (SUB-)DIRECTORY
  # check if current entry is a (sub-)directory ...
  if (-d $cur_entry_abs) {
    push @dirlist_path, $cur_entry;
    # ... and parse this (recursion here!!!)
    iterate_path($cur_entry_abs);
    pop @dirlist_path;
  }
}

sub iterate_path {

  my $path = shift;

  # create new "DirHandle"-object
  my $d = new DirHandle $path;
  if (defined $d) {

    # iterate through all entries in $path ($d->read) and call out entry-handler on each entry
    while (defined(my $line = $d->read)) { 
      entry_callback($line); 
    }

    undef $d;
  }
}

sub getDirList {

  $dirlist_base = shift;

  # reset vars
  $dirlist_buf = '';
  @dirlist_path = ();

  # start parsing file-structure
  iterate_path($dirlist_base);

  # return complete list of directory-content including files and subdirs
  # entries are newline (\n) - seperated
  return $dirlist_buf;

}
# =============================================


sub now {
  return strftime("%Y-%m-%d %H:%M:%S", localtime);
}

sub deep_copy {
  my $this = shift;
  if (not ref $this) {
    $this;
  } elsif (ref $this eq "ARRAY") {
    [map deep_copy($_), @$this];
  } elsif (ref $this eq "HASH") {
    +{map { $_ => deep_copy($this->{$_}) } keys %$this};
  } elsif (ref $this eq "CODE") {
    $this;
  } else { die "what type is $_?" }
}

1;
1	cvsjoko	1.1	#################################
2			#
3	joko	1.6	# $Id: libp.pm,v 1.5 2002/10/27 18:34:28 joko Exp $
4	cvsjoko	1.2	#
5			# $Log: libp.pm,v $
6	joko	1.6	# Revision 1.5 2002/10/27 18:34:28 joko
7			# + sub now
8			#
9	joko	1.5	# Revision 1.4 2002/08/16 19:06:39 cvsjoko
10			# + sub getDirList
11			#
12	cvsjoko	1.4	# Revision 1.3 2002/07/19 18:13:50 cvsjoko
13			# no message
14			#
15	cvsjoko	1.3	# Revision 1.2 2002/06/27 02:14:22 cvsjoko
16			# + stripHtml stripSpaces stripNewLines toReal
17			#
18	cvsjoko	1.2	# Revision 1.1 2002/06/24 14:49:59 cvsjoko
19			# + new
20	cvsjoko	1.1	#
21			#
22			#################################
23
24			package libp;
25
26			require Exporter;
27			@ISA = qw( Exporter );
28			@EXPORT = qw(
29	cvsjoko	1.3	Dumper
30			md5 md5_hex md5_base64
31			ParseDate UnixDate
32			strftime
33			stripHtml stripSpaces stripNewLines toReal trim
34			croak
35			array_getDifference
36	cvsjoko	1.4	getDirList
37	joko	1.5	now
38	joko	1.6	deep_copy
39	cvsjoko	1.1	);
40
41			use strict;
42			use warnings;
43
44			use Data::Dumper;
45			use Digest::MD5 qw(md5 md5_hex md5_base64);
46
47			$main::TZ = 'GMT';
48			use Date::Manip;
49	cvsjoko	1.2
50			require LWP::UserAgent;
51			use HTML::PullParser;
52
53	cvsjoko	1.3	# $now_string = strftime "%a %b %e %H:%M:%S %Y", localtime;
54			# see "perldoc -f localtime"
55			use POSIX qw(strftime);
56
57			use Carp;
58
59	cvsjoko	1.4	use DirHandle;
60
61	cvsjoko	1.2
62			########################################
63
64			sub stripSpaces {
65			my $text = shift;
66			#print "text: $text", "\n";
67			#print "ord: ", ord(substr($text, 0, 1)), "\n";
68			$text =~ s/^\s*//g;
69			$text =~ s/\s*$//g;
70			return $text;
71			}
72
73	cvsjoko	1.3	sub trim {
74			my $string = shift;
75			return stripSpaces($string);
76			}
77
78	cvsjoko	1.2	sub stripNewLines {
79			my $text = shift;
80			#print "text: $text", "\n";
81			#print "ord: ", ord(substr($text, 0, 1)), "\n";
82			$text =~ s/\n//g;
83			#$text =~ s/\s*$//g;
84			return $text;
85			}
86
87			sub toReal {
88			my $string = shift;
89			$string =~ m/(\d+\.*\d+)/;
90			my $real = $1;
91			return $real;
92			}
93
94			sub stripHtml {
95			my $html = shift;
96			my $result = '';
97			#$html =~ s/<br>(.*)/ - ($1)/i;
98			my $p = HTML::PullParser->new(
99			doc => \$html,
100			text => 'text',
101			unbroken_text => 1,
102			);
103			while (my $token = $p->get_token()) {
104			my $text = join('', @{$token});
105			$result .= $text;
106			}
107			#$result =~ s/ //g;
108			return $result;
109	cvsjoko	1.3	}
110
111			sub array_getRelations {
112			my $a_ref = shift;
113			my $b_ref = shift;
114			my @a = @{$a_ref};
115			my @b = @{$b_ref};
116
117			my @isect = my @diff = my @union = ();
118			my $e;
119			my %count;
120
121			foreach $e (@a, @b) { $count{$e}++ }
122
123			foreach $e (keys %count) {
124			push(@union, $e);
125			push @{ $count{$e} == 2 ? \@isect : \@diff }, $e;
126			}
127
128			my $result = {
129			union => \@union,
130			isect => \@isect,
131			diff => \@diff,
132			};
133
134			}
135
136			sub array_getDifference {
137			my $res = array_getRelations(shift, shift);
138			return $res->{diff};
139	cvsjoko	1.2	}
140	cvsjoko	1.4
141
142			# =============================================
143			# "global" vars used in directory-recursive-parsing
144			my $dirlist_buf;
145			my @dirlist_path;
146			my $dirlist_base;
147
148			sub entry_callback {
149
150			my $entry = shift;
151
152			# CHECKS
153			# dont't use this:
154			if ($entry eq '.' \|\| $entry eq '..') { return; }
155
156			# PREPARE
157			# prepare path to current entry
158			my $cur_entry = join('/', @dirlist_path, $entry);
159			# prepare path to current entry (absolute)
160			my $cur_entry_abs = join('/', $dirlist_base, @dirlist_path, $entry);
161
162			# ENTRY
163			# add current entry to buffer
164			$dirlist_buf .= $cur_entry . "\n";
165
166			# (SUB-)DIRECTORY
167			# check if current entry is a (sub-)directory ...
168			if (-d $cur_entry_abs) {
169			push @dirlist_path, $cur_entry;
170			# ... and parse this (recursion here!!!)
171			iterate_path($cur_entry_abs);
172			pop @dirlist_path;
173			}
174			}
175
176			sub iterate_path {
177
178			my $path = shift;
179
180			# create new "DirHandle"-object
181			my $d = new DirHandle $path;
182			if (defined $d) {
183
184			# iterate through all entries in $path ($d->read) and call out entry-handler on each entry
185			while (defined(my $line = $d->read)) {
186			entry_callback($line);
187			}
188
189			undef $d;
190			}
191			}
192
193			sub getDirList {
194
195			$dirlist_base = shift;
196
197			# reset vars
198			$dirlist_buf = '';
199			@dirlist_path = ();
200
201			# start parsing file-structure
202			iterate_path($dirlist_base);
203
204			# return complete list of directory-content including files and subdirs
205			# entries are newline (\n) - seperated
206			return $dirlist_buf;
207
208			}
209			# =============================================
210
211	joko	1.5
212			sub now {
213			return strftime("%Y-%m-%d %H:%M:%S", localtime);
214	joko	1.6	}
215
216			sub deep_copy {
217			my $this = shift;
218			if (not ref $this) {
219			$this;
220			} elsif (ref $this eq "ARRAY") {
221			[map deep_copy($_), @$this];
222			} elsif (ref $this eq "HASH") {
223			+{map { $_ => deep_copy($this->{$_}) } keys %$this};
224			} elsif (ref $this eq "CODE") {
225			$this;
226			} else { die "what type is $_?" }
227	joko	1.5	}
228	cvsjoko	1.1
229			1;
MailToCvsAdmin">MailToCvsAdmin	ViewVC Help
Powered by ViewVC 1.1.26