111 |
warn "prefixes found = ",dump($p); |
warn "prefixes found = ",dump($p); |
112 |
my @sorted = sort { $p->{$b} <=> $p->{$a} } keys %$p; |
my @sorted = sort { $p->{$b} <=> $p->{$a} } keys %$p; |
113 |
my $strip = shift @sorted || return @_; |
my $strip = shift @sorted || return @_; |
114 |
warn "longest preffix: '$strip'\n"; |
warn "longest prefix: '$strip' (stripped)\n"; |
115 |
return map { my $v = $_; $v =~ s/^\Q$strip\E//i; $v; } @_; |
return map { my $v = $_; $v =~ s/^\Q$strip\E//i; $v; } @_; |
116 |
} |
} |
117 |
|
|
118 |
|
sub group_by { |
119 |
|
my ( $group, $detail ) = ( shift, shift ); |
120 |
|
my $what = shift; |
121 |
|
confess "expected CODE as first argument!" unless ref($what) eq 'CODE'; |
122 |
|
my @data = @_; |
123 |
|
|
124 |
|
my $stat; |
125 |
|
my @details; |
126 |
|
|
127 |
|
foreach my $i ( 0 .. $#data ) { |
128 |
|
my $v = $data[$i]; |
129 |
|
my ( $by, $rest ) = $what->($v); |
130 |
|
# warn "## group_by: $i $v -> $by\n"; |
131 |
|
push @{ $stat->{$by} }, $i; |
132 |
|
$details[$i] = $rest; |
133 |
|
} |
134 |
|
|
135 |
|
warn "group_by '$group' stats = ",dump( $stat ); |
136 |
|
|
137 |
|
foreach my $g ( keys %$stat ) { |
138 |
|
foreach my $r ( @{ $stat->{$g} } ) { |
139 |
|
warn "## $group $g $r\n"; |
140 |
|
$dump[$r]->{$group} = $g; |
141 |
|
$dump[$r]->{$detail} = $details[$r]; |
142 |
|
} |
143 |
|
} |
144 |
|
} |
145 |
|
|
146 |
while (<$csv_fh>) { |
while (<$csv_fh>) { |
147 |
$csv_parser->parse($_); |
$csv_parser->parse($_); |
148 |
my @fields = $csv_parser->fields; |
my @fields = $csv_parser->fields; |
173 |
|
|
174 |
my $v = clean( $sv[$j] ); |
my $v = clean( $sv[$j] ); |
175 |
|
|
|
if ( $j == 0 ) { |
|
|
$h->{ $l . '_short' } = $v; |
|
|
} |
|
|
|
|
176 |
if ( my $human = human( $v ) ) { |
if ( my $human = human( $v ) ) { |
177 |
$h->{ $l . '_' . $j . '_human' } = $human; |
$h->{ $l . '_' . $j . '_human' } = $human; |
178 |
} else { |
} else { |
225 |
|
|
226 |
warn "split_stats = ", dump( $split_stats ), "\n"; |
warn "split_stats = ", dump( $split_stats ), "\n"; |
227 |
|
|
228 |
|
# |
229 |
|
# split prefix from label_0 |
230 |
|
# |
231 |
my @all = map { $_->{label_0} || die "no label_0 for ",dump($_) } @dump; |
my @all = map { $_->{label_0} || die "no label_0 for ",dump($_) } @dump; |
232 |
warn "all = ",dump(@all); |
warn "all = ",dump(@all); |
233 |
my @stripped = strip_prefix( @all ); |
my @stripped = strip_prefix( @all ); |
234 |
$dump[$_]->{label_0} = $stripped[$_] foreach ( 0 .. $#stripped ); |
$dump[$_]->{label_0} = $stripped[$_] foreach ( 0 .. $#stripped ); |
235 |
|
|
236 |
|
# group products by manufacturers |
237 |
|
group_by( qw/manufacturer player_name/, sub { $_[0] =~ m/^(\S+)\s+(.+)/; ($1,$2) }, @stripped ); |
238 |
|
|
239 |
warn "dump = ", dump( @dump ), "\n"; |
warn "dump = ", dump( @dump ), "\n"; |
240 |
|
|
241 |
print "features: .", join(', .', keys %$split_stats), "\n"; |
print "features: .", join(', .', keys %$split_stats), "\n"; |