document html table format in directory.html
[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
12 __PACKAGE__->attr('path');
13 __PACKAGE__->attr('full_path');
14
15 sub data {
16         my $self = shift;
17
18         my $path = $self->path;
19
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);
24         my @header;
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 => [] };
31
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 ) {
37                         $line =~ s/\^//g;
38                         chomp $line;
39                         my @v = split(/\|/, $line);
40                         while ( @lines && $#v < $#header ) {
41                                 $line = $lines[0];
42                                 $line =~ s/\^//g;
43                                 chomp $line;
44                                 my @more_v = split(/\|/, $line);
45                                 if ( $#v + $#more_v > $#header ) {
46                                         warn "short line: ",dump( @v );
47                                         last;
48                                 }
49                                 shift @lines;
50                                 $v[ $#v ] .= shift @more_v if @more_v;
51                                 push @v, @more_v if @more_v;
52
53                                 if ( $#v > $#header ) {
54                                         die "# splice $#header ", dump( @v );
55                                         @v = splice @v, 0, $#header;
56                                 }
57                         }
58                         my $item;
59                         foreach my $i ( 0 .. $#v ) {
60                                 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
61                         }
62                         push @{ $data->{items} }, $item;
63                 }
64         } else {
65                 warn "file format unknown $path";
66         }
67
68         $data->{header} = [ @header ];
69         
70         return $data;
71
72 }
73
74 1