line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
1
|
|
|
|
|
|
|
package DBIx::TextIndex::TermDocsCache; |
2
|
|
|
|
|
|
|
|
3
|
8
|
|
|
8
|
|
51
|
use strict; |
|
8
|
|
|
|
|
15
|
|
|
8
|
|
|
|
|
323
|
|
4
|
8
|
|
|
8
|
|
45
|
use warnings; |
|
8
|
|
|
|
|
15
|
|
|
8
|
|
|
|
|
484
|
|
5
|
|
|
|
|
|
|
|
6
|
|
|
|
|
|
|
our $VERSION = '0.26'; |
7
|
|
|
|
|
|
|
|
8
|
8
|
|
|
8
|
|
49
|
use Bit::Vector; |
|
8
|
|
|
|
|
13
|
|
|
8
|
|
|
|
|
409
|
|
9
|
|
|
|
|
|
|
|
10
|
8
|
|
|
8
|
|
45
|
use base qw(DBIx::TextIndex); |
|
8
|
|
|
|
|
17
|
|
|
8
|
|
|
|
|
5361
|
|
11
|
|
|
|
|
|
|
|
12
|
|
|
|
|
|
|
sub new { |
13
|
0
|
|
|
0
|
1
|
|
my $pkg = shift; |
14
|
0
|
|
0
|
|
|
|
my $class = ref($pkg) || $pkg; |
15
|
0
|
|
|
|
|
|
my $self = bless {}, $class; |
16
|
0
|
|
|
|
|
|
my $args = shift; |
17
|
0
|
0
|
|
|
|
|
$self->set($args) if $args; |
18
|
0
|
|
|
|
|
|
return $self; |
19
|
|
|
|
|
|
|
} |
20
|
|
|
|
|
|
|
|
21
|
|
|
|
|
|
|
sub max_indexed_id { |
22
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
23
|
0
|
0
|
|
|
|
|
if (@_) { |
24
|
0
|
|
|
|
|
|
$self->flush_all; |
25
|
0
|
|
|
|
|
|
$self->{MAX_INDEXED_ID} = $_[0]; |
26
|
|
|
|
|
|
|
} |
27
|
0
|
|
|
|
|
|
return $self->{MAX_INDEXED_ID}; |
28
|
|
|
|
|
|
|
} |
29
|
|
|
|
|
|
|
|
30
|
|
|
|
|
|
|
sub flush_all { |
31
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
32
|
0
|
|
|
|
|
|
$self->flush_bit_vectors; |
33
|
0
|
|
|
|
|
|
$self->flush_term_docs; |
34
|
|
|
|
|
|
|
} |
35
|
|
|
|
|
|
|
|
36
|
|
|
|
|
|
|
sub flush_bit_vectors { |
37
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
38
|
0
|
|
|
|
|
|
delete($self->{VECTOR}); |
39
|
|
|
|
|
|
|
} |
40
|
|
|
|
|
|
|
|
41
|
|
|
|
|
|
|
sub flush_term_docs { |
42
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
43
|
0
|
|
|
|
|
|
delete($self->{TERM_DOCS}); |
44
|
0
|
|
|
|
|
|
delete($self->{DOCFREQ_T}); |
45
|
|
|
|
|
|
|
} |
46
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
sub term_pos { |
48
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
49
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
50
|
0
|
0
|
|
|
|
|
$self->_fetch_term_pos($fno, $term) unless exists $self->{TERM_POS}->[$fno]->{$term}; |
51
|
0
|
|
|
|
|
|
return $self->{TERM_POS}->[$fno]->{$term}; |
52
|
|
|
|
|
|
|
} |
53
|
|
|
|
|
|
|
|
54
|
|
|
|
|
|
|
sub term_docs { |
55
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
56
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
57
|
0
|
0
|
|
|
|
|
$self->_fetch_term_docs($fno, $term) unless exists $self->{TERM_DOCS}->[$fno]->{$term}; |
58
|
0
|
|
|
|
|
|
return $self->{TERM_DOCS}->[$fno]->{$term}; |
59
|
|
|
|
|
|
|
} |
60
|
|
|
|
|
|
|
|
61
|
|
|
|
|
|
|
sub term_docs_hashref { |
62
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
63
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
64
|
0
|
0
|
|
|
|
|
$self->_fetch_term_docs($fno, $term) unless exists $self->{TERM_DOCS}->[$fno]->{$term}; |
65
|
0
|
|
|
|
|
|
return DBIx::TextIndex::term_docs_hashref($self->{TERM_DOCS}->[$fno]->{$term}); |
66
|
|
|
|
|
|
|
|
67
|
|
|
|
|
|
|
} |
68
|
|
|
|
|
|
|
|
69
|
|
|
|
|
|
|
sub term_docs_arrayref { |
70
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
71
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
72
|
0
|
0
|
|
|
|
|
$self->_fetch_term_docs($fno, $term) unless exists $self->{TERM_DOCS}->[$fno]->{$term}; |
73
|
0
|
|
|
|
|
|
return DBIx::TextIndex::term_docs_arrayref($self->{TERM_DOCS}->[$fno]->{$term}); |
74
|
|
|
|
|
|
|
} |
75
|
|
|
|
|
|
|
|
76
|
|
|
|
|
|
|
sub term_doc_ids_arrayref { |
77
|
8
|
|
|
8
|
|
53
|
no warnings qw(uninitialized); |
|
8
|
|
|
|
|
13
|
|
|
8
|
|
|
|
|
4395
|
|
78
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
79
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
80
|
0
|
0
|
|
|
|
|
$self->_fetch_term_docs($fno, $term) unless exists $self->{TERM_DOCS}->[$fno]->{$term}; |
81
|
0
|
|
|
|
|
|
return DBIx::TextIndex::term_doc_ids_arrayref($self->{TERM_DOCS}->[$fno]->{$term}); |
82
|
|
|
|
|
|
|
} |
83
|
|
|
|
|
|
|
|
84
|
|
|
|
|
|
|
sub vector { |
85
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
86
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
87
|
0
|
0
|
|
|
|
|
if ($self->{VECTOR}->[$fno]->{$term}) { |
88
|
0
|
|
|
|
|
|
return $self->{VECTOR}->[$fno]->{$term}; |
89
|
|
|
|
|
|
|
} |
90
|
0
|
|
|
|
|
|
my $doc_ids = $self->term_doc_ids_arrayref($fno, $term); |
91
|
0
|
|
|
|
|
|
my $vector = Bit::Vector->new($self->{MAX_INDEXED_ID} + 1); |
92
|
0
|
|
|
|
|
|
$vector->Index_List_Store(@$doc_ids); |
93
|
0
|
|
|
|
|
|
$self->{VECTOR}->[$fno]->{$term} = $vector; |
94
|
0
|
|
|
|
|
|
return $vector; |
95
|
|
|
|
|
|
|
} |
96
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
sub f_t { |
98
|
0
|
|
|
0
|
1
|
|
my $self = shift; |
99
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
100
|
0
|
0
|
|
|
|
|
$self->_fetch_term_docs($fno, $term) |
101
|
|
|
|
|
|
|
unless exists $self->{DOCFREQ_T}->[$fno]->{$term}; |
102
|
0
|
|
|
|
|
|
return $self->{DOCFREQ_T}->[$fno]->{$term}; |
103
|
|
|
|
|
|
|
} |
104
|
|
|
|
|
|
|
|
105
|
|
|
|
|
|
|
sub _fetch_term_pos { |
106
|
0
|
|
|
0
|
|
|
my $self = shift; |
107
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
108
|
0
|
|
|
|
|
|
my $sql = |
109
|
|
|
|
|
|
|
$self->{DB}->fetch_term_pos($self->{INVERTED_TABLES}->[$fno]); |
110
|
|
|
|
|
|
|
|
111
|
0
|
|
|
|
|
|
($self->{TERM_POS}->[$fno]->{$term}) |
112
|
|
|
|
|
|
|
= $self->{INDEX_DBH}->selectrow_array($sql, undef, $term); |
113
|
|
|
|
|
|
|
} |
114
|
|
|
|
|
|
|
|
115
|
|
|
|
|
|
|
sub _fetch_term_docs { |
116
|
0
|
|
|
0
|
|
|
my $self = shift; |
117
|
0
|
|
|
|
|
|
my ($fno, $term) = @_; |
118
|
0
|
|
|
|
|
|
my $sql = $self->{DB}->fetch_term_freq_and_docs( |
119
|
|
|
|
|
|
|
$self->{INVERTED_TABLES}->[$fno]); |
120
|
|
|
|
|
|
|
|
121
|
0
|
|
|
|
|
|
($self->{DOCFREQ_T}->[$fno]->{$term}, $self->{TERM_DOCS}->[$fno]->{$term}) |
122
|
|
|
|
|
|
|
= $self->{INDEX_DBH}->selectrow_array($sql, undef, $term); |
123
|
|
|
|
|
|
|
} |
124
|
|
|
|
|
|
|
|
125
|
|
|
|
|
|
|
__END__ |