fix json loading
[MojoFacets.git] / lib / MojoFacets / Import / File.pm
1 package MojoFacets::Import::File;
2
3 use warnings;
4 use strict;
5
6 use base 'Mojo::Base';
7
8 use HTML::TableExtract;
9 use File::Slurp;
10 use Data::Dump qw(dump);
11 use JSON;
12
13 __PACKAGE__->attr('path');
14 __PACKAGE__->attr('full_path');
15
16 sub data {
17         my $self = shift;
18
19         my $path = $self->path;
20
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);
25         my @header;
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 => [] };
32
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 ) {
38                         $line =~ s/\^//g;
39                         chomp $line;
40                         my @v = split(/\|/, $line);
41                         while ( @lines && $#v < $#header ) {
42                                 $line = $lines[0];
43                                 $line =~ s/\^//g;
44                                 chomp $line;
45                                 my @more_v = split(/\|/, $line);
46                                 if ( $#v + $#more_v > $#header ) {
47                                         warn "short line: ",dump( @v );
48                                         last;
49                                 }
50                                 shift @lines;
51                                 $v[ $#v ] .= shift @more_v if @more_v;
52                                 push @v, @more_v if @more_v;
53
54                                 if ( $#v > $#header ) {
55                                         die "# splice $#header ", dump( @v );
56                                         @v = splice @v, 0, $#header;
57                                 }
58                         }
59                         my $item;
60                         foreach my $i ( 0 .. $#v ) {
61                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
62                         }
63                         push @{ $data->{items} }, $item;
64                 }
65         } else {
66                 warn "file format unknown $path";
67         }
68
69         $data->{header} = [ @header ];
70         
71         return $data;
72
73 }
74
75 1