1 package MojoFacets::Import::File;
8 use HTML::TableExtract;
10 use Data::Dump qw(dump);
13 __PACKAGE__->attr('path');
14 __PACKAGE__->attr('full_path');
19 my $path = $self->path;
21 # we could use Mojo::JSON here, but it's too slow
22 # $data = from_json read_file $path;
23 my $data = read_file $self->full_path;
24 warn "# data snippet: ", substr($data,0,200);
26 if ( $path =~ m/\.js/ ) {
27 Encode::_utf8_on($data);
28 $data = from_json $data;
29 } elsif ( $path =~ m/\.txt/ ) {
30 my @lines = split(/\r?\n/, $data);
31 $data = { items => [] };
33 my $header_line = shift @lines;
34 my $multiline = $header_line =~ s/\^//g;
35 @header = split(/\|/, $header_line );
36 warn "# header ", dump( @header );
37 while ( my $line = shift @lines ) {
40 my @v = split(/\|/, $line);
41 while ( @lines && $#v < $#header ) {
45 my @more_v = split(/\|/, $line);
46 if ( $#v + $#more_v > $#header ) {
47 warn "short line: ",dump( @v );
51 $v[ $#v ] .= shift @more_v if @more_v;
52 push @v, @more_v if @more_v;
54 if ( $#v > $#header ) {
55 die "# splice $#header ", dump( @v );
56 @v = splice @v, 0, $#header;
60 foreach my $i ( 0 .. $#v ) {
61 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
63 push @{ $data->{items} }, $item;
66 warn "file format unknown $path";
69 $data->{header} = [ @header ];