503be1d59c61179230afb32d6387995f55199890
[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 ext { '\.(js(on)?|txt)$' }
17
18 sub data {
19         my $self = shift;
20
21         my $path = $self->path;
22
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);
27         my @header;
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 => [] };
34
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 ) {
40                         $line =~ s/\^//g;
41                         chomp $line;
42                         my @v = split(/\|/, $line);
43                         while ( @lines && $#v < $#header ) {
44                                 $line = $lines[0];
45                                 $line =~ s/\^//g;
46                                 chomp $line;
47                                 my @more_v = split(/\|/, $line);
48                                 if ( $#v + $#more_v > $#header ) {
49                                         warn "short line: ",dump( @v );
50                                         last;
51                                 }
52                                 shift @lines;
53                                 $v[ $#v ] .= shift @more_v if @more_v;
54                                 push @v, @more_v if @more_v;
55
56                                 if ( $#v > $#header ) {
57                                         die "# splice $#header ", dump( @v );
58                                         @v = splice @v, 0, $#header;
59                                 }
60                         }
61                         my $item;
62                         foreach my $i ( 0 .. $#v ) {
63                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
64                         }
65                         push @{ $data->{items} }, $item;
66                 }
67         } else {
68                 warn "file format unknown $path";
69         }
70
71         $data->{header} = [ @header ];
72         
73         return $data;
74
75 }
76
77 1