1 package MojoFacets::Import::File;
8 use HTML::TableExtract;
10 use Data::Dump qw(dump);
13 __PACKAGE__->attr('path');
14 __PACKAGE__->attr('full_path');
16 sub ext { '\.(js(on)?|txt)$' }
21 my $path = $self->path;
23 # we could use Mojo::JSON here, but it's too slow
24 # $data = from_json read_file $path;
25 my $data = read_file $self->full_path;
26 warn "# data snippet: ", substr($data,0,200);
28 if ( $path =~ m/\.js(on)?/ ) {
29 Encode::_utf8_on($data);
30 $data = from_json $data;
31 } elsif ( $path =~ m/\.txt/ ) {
32 my @lines = split(/\r?\n/, $data);
33 $data = { items => [] };
35 my $header_line = shift @lines;
36 my $multiline = $header_line =~ s/\^//g;
37 @header = split(/\|/, $header_line );
38 warn "# header ", dump( @header );
39 while ( my $line = shift @lines ) {
42 my @v = split(/\|/, $line);
43 while ( @lines && $#v < $#header ) {
47 my @more_v = split(/\|/, $line);
48 if ( $#v + $#more_v > $#header ) {
49 warn "short line: ",dump( @v );
53 $v[ $#v ] .= shift @more_v if @more_v;
54 push @v, @more_v if @more_v;
56 if ( $#v > $#header ) {
57 die "# splice $#header ", dump( @v );
58 @v = splice @v, 0, $#header;
62 foreach my $i ( 0 .. $#v ) {
63 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
65 push @{ $data->{items} }, $item;
68 warn "file format unknown $path";
71 $data->{header} = [ @header ];