1 package MojoFacets::Import::File;
8 use HTML::TableExtract;
10 use Data::Dump qw(dump);
12 __PACKAGE__->attr('path');
13 __PACKAGE__->attr('full_path');
18 my $path = $self->path;
20 # we could use Mojo::JSON here, but it's too slow
21 # $data = from_json read_file $path;
22 my $data = read_file $self->path_path;
23 warn "# data snippet: ", substr($data,0,200);
25 if ( $path =~ m/\.js/ ) {
26 Encode::_utf8_on($data);
27 $data = from_json $data;
28 } elsif ( $path =~ m/\.txt/ ) {
29 my @lines = split(/\r?\n/, $data);
30 $data = { items => [] };
32 my $header_line = shift @lines;
33 my $multiline = $header_line =~ s/\^//g;
34 @header = split(/\|/, $header_line );
35 warn "# header ", dump( @header );
36 while ( my $line = shift @lines ) {
39 my @v = split(/\|/, $line);
40 while ( @lines && $#v < $#header ) {
44 my @more_v = split(/\|/, $line);
45 if ( $#v + $#more_v > $#header ) {
46 warn "short line: ",dump( @v );
50 $v[ $#v ] .= shift @more_v if @more_v;
51 push @v, @more_v if @more_v;
53 if ( $#v > $#header ) {
54 die "# splice $#header ", dump( @v );
55 @v = splice @v, 0, $#header;
59 foreach my $i ( 0 .. $#v ) {
60 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
62 push @{ $data->{items} }, $item;
65 warn "file format unknown $path";
68 $data->{header} = [ @header ];