summaryrefslogtreecommitdiff
path: root/Master/tlpkg/tlperl/lib/Tie/Memoize.pm
blob: 2793a04590d46f1299a47551ce3d272bec65d011 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
use strict;
package Tie::Memoize;
use Tie::Hash;
our @ISA = 'Tie::ExtraHash';
our $VERSION = '1.0';

our $exists_token = \undef;

sub croak {require Carp; goto &Carp::croak}

# Format: [0: STORAGE, 1: EXISTS-CACHE, 2: FETCH_function;
#	   3: EXISTS_function, 4: DATA, 5: EXISTS_different ]

sub FETCH {
  my ($h,$key) = ($_[0][0], $_[1]);
  my $res = $h->{$key};
  return $res if defined $res;	# Shortcut if accessible
  return $res if exists $h->{$key}; # Accessible, but undef
  my $cache = $_[0][1]{$key};
  return if defined $cache and not $cache; # Known to not exist
  my @res = $_[0][2]->($key, $_[0][4]);	# Autoload
  $_[0][1]{$key} = 0, return unless @res; # Cache non-existence
  delete $_[0][1]{$key};	# Clear existence cache, not needed any more
  $_[0][0]{$key} = $res[0];	# Store data and return
}

sub EXISTS   {
  my ($a,$key) = (shift, shift);
  return 1 if exists $a->[0]{$key}; # Have data
  my $cache = $a->[1]{$key};
  return $cache if defined $cache; # Existence cache
  my @res = $a->[3]($key,$a->[4]);
  $_[0][1]{$key} = 0, return unless @res; # Cache non-existence
  # Now we know it exists
  return ($_[0][1]{$key} = 1) if $a->[5]; # Only existence reported
  # Now know the value
  $_[0][0]{$key} = $res[0];	# Store data
  return 1
}

sub TIEHASH  {
  croak 'syntax: tie %hash, \'Tie::AutoLoad\', \&fetch_subr' if @_ < 2;
  croak 'syntax: tie %hash, \'Tie::AutoLoad\', \&fetch_subr, $data, \&exists_subr, \%data_cache, \%existence_cache' if @_ > 6;
  push @_, undef if @_ < 3;	# Data
  push @_, $_[1] if @_ < 4;	# exists
  push @_, {} while @_ < 6;	# initial value and caches
  bless [ @_[4,5,1,3,2], $_[1] ne $_[3]], $_[0]
}

1;

=head1 NAME

Tie::Memoize - add data to hash when needed

=head1 SYNOPSIS

  require Tie::Memoize;
  tie %hash, 'Tie::Memoize',
      \&fetch,			# The rest is optional
      $DATA, \&exists,
      {%ini_value}, {%ini_existence};

=head1 DESCRIPTION

This package allows a tied hash to autoload its values on the first access,
and to use the cached value on the following accesses.

Only read-accesses (via fetching the value or C<exists>) result in calls to
the functions; the modify-accesses are performed as on a normal hash.

The required arguments during C<tie> are the hash, the package, and
the reference to the C<FETCH>ing function.  The optional arguments are
an arbitrary scalar $data, the reference to the C<EXISTS> function,
and initial values of the hash and of the existence cache.

Both the C<FETCH>ing function and the C<EXISTS> functions have the
same signature: the arguments are C<$key, $data>; $data is the same
value as given as argument during tie()ing.  Both functions should
return an empty list if the value does not exist.  If C<EXISTS>
function is different from the C<FETCH>ing function, it should return
a TRUE value on success.  The C<FETCH>ing function should return the
intended value if the key is valid.

=head1 Inheriting from B<Tie::Memoize>

The structure of the tied() data is an array reference with elements

  0:  cache of known values
  1:  cache of known existence of keys
  2:  FETCH  function
  3:  EXISTS function
  4:  $data

The rest is for internal usage of this package.  In particular, if
TIEHASH is overwritten, it should call SUPER::TIEHASH.

=head1 EXAMPLE

  sub slurp {
    my ($key, $dir) = shift;
    open my $h, '<', "$dir/$key" or return;
    local $/; <$h>			# slurp it all
  }
  sub exists { my ($key, $dir) = shift; return -f "$dir/$key" }

  tie %hash, 'Tie::Memoize', \&slurp, $directory, \&exists,
      { fake_file1 => $content1, fake_file2 => $content2 },
      { pretend_does_not_exists => 0, known_to_exist => 1 };

This example treats the slightly modified contents of $directory as a
hash.  The modifications are that the keys F<fake_file1> and
F<fake_file2> fetch values $content1 and $content2, and
F<pretend_does_not_exists> will never be accessed.  Additionally, the
existence of F<known_to_exist> is never checked (so if it does not
exists when its content is needed, the user of %hash may be confused).

=head1 BUGS

FIRSTKEY and NEXTKEY methods go through the keys which were already read,
not all the possible keys of the hash.

=head1 AUTHOR

Ilya Zakharevich L<mailto:perl-module-hash-memoize@ilyaz.org>.

=cut