| 1 |
wakaba |
1.1 |
package SWE::Object::Document; |
| 2 |
|
|
use strict; |
| 3 |
|
|
use warnings; |
| 4 |
|
|
|
| 5 |
|
|
sub new ($%) { |
| 6 |
|
|
my $class = shift; |
| 7 |
|
|
my $self = bless {@_}, $class; |
| 8 |
|
|
|
| 9 |
|
|
return $self; |
| 10 |
|
|
} |
| 11 |
|
|
|
| 12 |
wakaba |
1.3 |
sub new_id ($%) { |
| 13 |
|
|
my $self = shift->new (@_); |
| 14 |
|
|
|
| 15 |
|
|
my $idgen = $self->db->id; |
| 16 |
|
|
my $id = $idgen->get_next_id; |
| 17 |
|
|
$self->{id} = $id; |
| 18 |
|
|
|
| 19 |
|
|
return $self; |
| 20 |
|
|
} # new_id |
| 21 |
|
|
|
| 22 |
wakaba |
1.1 |
sub db { $_[0]->{db} } |
| 23 |
|
|
|
| 24 |
|
|
sub id { $_[0]->{id} } |
| 25 |
|
|
|
| 26 |
|
|
sub associate_names ($$%) { |
| 27 |
|
|
my ($self, $names, %args) = @_; |
| 28 |
|
|
|
| 29 |
|
|
## NOTE: names_lock MUST be executed before the invocation. |
| 30 |
|
|
|
| 31 |
|
|
my $id = $self->id; |
| 32 |
|
|
my $time = $args{time} || time; |
| 33 |
|
|
my $sw3_pages = $self->{sw3_pages}; ## TODO: ... |
| 34 |
|
|
|
| 35 |
|
|
my $vc = $self->db->vc; |
| 36 |
|
|
|
| 37 |
|
|
my $name_prop_db = $self->{name_prop_db}; ## TODO: ... |
| 38 |
|
|
local $name_prop_db->{version_control} = $vc; |
| 39 |
|
|
|
| 40 |
|
|
my $name_history_db = $self->db->name_history; |
| 41 |
|
|
local $name_history_db->{version_control} = $vc; |
| 42 |
|
|
|
| 43 |
|
|
for my $name (keys %$names) { |
| 44 |
|
|
my $name_props = $name_prop_db->get_data ($name); |
| 45 |
|
|
unless (defined $name_props) { |
| 46 |
|
|
my $sw3id = $sw3_pages->get_data ($name); |
| 47 |
|
|
main::convert_sw3_page ($sw3id => $name); ## TODO: ... |
| 48 |
|
|
|
| 49 |
|
|
$name_props = $name_prop_db->get_data ($name); |
| 50 |
|
|
unless (defined $name_props) { |
| 51 |
|
|
$name_history_db->append_data ($name => [$time, 'c']); |
| 52 |
|
|
} |
| 53 |
|
|
} |
| 54 |
|
|
|
| 55 |
|
|
push @{$name_props->{id} ||= []}, $id; |
| 56 |
|
|
$name_props->{name} = $name; |
| 57 |
|
|
$name_prop_db->set_data ($name => $name_props); |
| 58 |
|
|
|
| 59 |
|
|
$name_history_db->append_data ($name => [$time, 'a', $id]); |
| 60 |
|
|
} |
| 61 |
|
|
|
| 62 |
|
|
my $user = $args{user} || '(anon)'; |
| 63 |
|
|
$vc->commit_changes ("id=$id created by $user"); |
| 64 |
|
|
} # associate_names |
| 65 |
|
|
|
| 66 |
wakaba |
1.2 |
sub update_tfidf ($$) { |
| 67 |
|
|
my ($self, $doc) = @_; ## TODO: $doc should not be an argument |
| 68 |
|
|
|
| 69 |
|
|
## It is REQUIRED to lock the $id before the invocation of this |
| 70 |
|
|
## method to keep the consistency of tfidf data for the $id. |
| 71 |
|
|
|
| 72 |
|
|
my $id = $self->id; |
| 73 |
|
|
|
| 74 |
|
|
my $tfidf_db = $self->db->id_tfidf; |
| 75 |
|
|
|
| 76 |
|
|
require SWE::Data::FeatureVector; |
| 77 |
|
|
|
| 78 |
|
|
my $deleted_terms = SWE::Data::FeatureVector->parse_stringref |
| 79 |
|
|
($tfidf_db->get_data ($id))->as_key_hashref; |
| 80 |
|
|
|
| 81 |
|
|
my $tc = $doc->document_element->text_content; |
| 82 |
|
|
|
| 83 |
|
|
## TODO: use element semantics... |
| 84 |
|
|
|
| 85 |
|
|
my $orig_tfs = {}; |
| 86 |
|
|
my $all_terms = 0; |
| 87 |
|
|
main::for_unique_words ($tc => sub { |
| 88 |
|
|
$orig_tfs->{$_[0]} = $_[1]; |
| 89 |
|
|
$all_terms += $_[1]; |
| 90 |
|
|
}); ## TODO: XXX |
| 91 |
|
|
|
| 92 |
|
|
my $names_index_db = $self->db->name_inverted_index; |
| 93 |
|
|
$names_index_db->lock; |
| 94 |
|
|
|
| 95 |
|
|
my $idgen = $self->db->id; |
| 96 |
|
|
my $doc_number = $idgen->get_last_id; |
| 97 |
|
|
|
| 98 |
|
|
my $terms = SWE::Data::FeatureVector->new; |
| 99 |
|
|
for my $term (keys %$orig_tfs) { |
| 100 |
|
|
my $n_tf = $orig_tfs->{$term} / $all_terms; |
| 101 |
|
|
|
| 102 |
|
|
my $df = $names_index_db->get_count ($term); |
| 103 |
|
|
my $idf = log ($doc_number / ($df + 1)); |
| 104 |
|
|
|
| 105 |
|
|
my $tfidf = $n_tf * $idf; |
| 106 |
|
|
|
| 107 |
|
|
$terms->set_tfidf ($term, $tfidf); |
| 108 |
|
|
$names_index_db->add_data ($term => $id => $tfidf); |
| 109 |
|
|
|
| 110 |
|
|
delete $deleted_terms->{$term}; |
| 111 |
|
|
} |
| 112 |
|
|
|
| 113 |
|
|
for my $term (keys %$deleted_terms) { |
| 114 |
|
|
$names_index_db->delete_data ($term, $id); |
| 115 |
|
|
} |
| 116 |
|
|
|
| 117 |
|
|
$tfidf_db->set_data ($id => \( $terms->stringify )); |
| 118 |
|
|
} # update_tfidf |
| 119 |
|
|
|
| 120 |
wakaba |
1.1 |
1; |