Fixed bug in pre-ranking algorithm
Change-Id: I9ae430129859fd0e48ca55713243c71c87b238a2
diff --git a/lib/Krawfish/Index/Fields/Direction.pm b/lib/Krawfish/Index/Fields/Direction.pm
index fa999c4..91e38e3 100644
--- a/lib/Krawfish/Index/Fields/Direction.pm
+++ b/lib/Krawfish/Index/Fields/Direction.pm
@@ -3,6 +3,8 @@
use strict;
use warnings;
+use constant DEBUG => 1;
+
sub new {
my $class = shift;
my $list = shift;
@@ -11,6 +13,14 @@
sub rank_for {
my ($self, $doc_id) = @_;
+
+ if (DEBUG) {
+ print_log(
+ 'f_rank_dir',
+ 'Get rank for doc ' . $doc_id . ' which is ' . $self->[$doc_id]
+ );
+ };
+
return $self->[$doc_id] // 0;
};
diff --git a/lib/Krawfish/Index/Fields/Rank.pm b/lib/Krawfish/Index/Fields/Rank.pm
index 7b7e97f..4dcca72 100644
--- a/lib/Krawfish/Index/Fields/Rank.pm
+++ b/lib/Krawfish/Index/Fields/Rank.pm
@@ -43,7 +43,7 @@
my ($value, $doc_id) = @_;
if (DEBUG) {
- print_log('f_rank', qq!Add value "$value" associated to ! . $doc_id)
+ print_log('f_rank', qq!Add value "$value" associated to $doc_id!)
};
# Collation is numerical
@@ -85,29 +85,46 @@
if (DEBUG) {
print_log(
'f_rank',
- 'Presorted list is ' .
+ 'Presorted list [VALUE,DOC] is ' .
join('', map { '[' . join(',',@$_) . ']' } @presort)
);
};
- # Remove duplicates
- my @sort;
- my $last_value;
- foreach my $next (@presort) {
- if ($last_value && $next->[0] eq $last_value) {
- push @{$sort[-1]}, $next->[1]
- }
- else {
- push @sort, [$next->[1]];
- $last_value = $next->[1];
- };
- };
-
-
# This list keeps existing, even
# when the segment becomes static -
# to make merging possible without the
# need to reconsult the dictionary
+
+ # Remove duplicates
+ my @sort;
+ my $last_value;
+
+ # Iterate over presort list
+ foreach my $next (@presort) {
+
+ # The last value is given and it's equal to the next value
+ if (defined $last_value && ($next->[0] eq $last_value)) {
+
+ # Add to the last added list
+ push @{$sort[-1]}, $next->[1]
+ }
+ else {
+
+ # Create new item
+ push @sort, [$next->[1]];
+ $last_value = $next->[0];
+ };
+ };
+
+ if (DEBUG) {
+ print_log(
+ 'f_rank',
+ 'Sorted list on ranks [DOC*] is ' .
+ join('', map { '[' . join(',',@$_) . ']' } @sort)
+ );
+ };
+
+
$self->{sorted} = \@sort;
# Create the ascending rank
@@ -127,6 +144,14 @@
$rank++;
};
+ if (DEBUG) {
+ print_log(
+ 'f_rank',
+ 'Ascending ranks per doc are ' .
+ join('', map { "[${_}]" } @asc)
+ );
+ };
+
$self->{asc} = Krawfish::Index::Fields::Direction->new(\@asc);
# Max rank is relevant for efficient encoding