1 package MojoFacets::Import::CSV;
9 use Data::Dump qw(dump);
12 __PACKAGE__->attr('path');
13 __PACKAGE__->attr('full_path'); # FIXME remove full_path
15 my $null = ''; # FIXME undef?
18 my ( $delimiter, $line ) = @_;
22 $line =~ s/""/_qq_/gc;
23 if ( $line =~ s/^"\s*([^"]+)\s*"\Q$delimiter\E?// ) {
25 } elsif ( $line =~ s/^\s*([^\Q$delimiter\E]+)\s*\Q$delimiter\E?// ) {
27 } elsif ( $line =~ s/^\s*\Q$delimiter\E// ) {
30 die "can't parse [$line]\n";
33 $v =~ s/^\s*(.+?)\s*$/$1/;
44 my $path = $self->full_path || $self->path;
46 my $data = read_file $path, { binmode => ':raw' }; # FIXME configurable!
47 my $encoding = 'utf-8';
48 if ( $path =~ m/\.(\w+).csv/i ) {
51 warn "decoding ", length($data), " bytes using $encoding\n";
52 $data = decode($encoding, $data);
54 my @lines = split(/\r?\n/, $data);
55 $data = { items => [] };
59 if ( $lines[0] !~ /;/ && $lines[1] =~ /;/ ) {
60 shift @lines; # FIXME ship non-header line
64 warn "$path ", $#lines + 1, " lines encoding: $encoding delimiter:",dump($delimiter);
66 my $header_line = shift @lines;
68 my @header = _split_line( $delimiter, $header_line );
69 warn "# header ",dump( @header );
71 while ( my $line = shift @lines ) {
73 my @v = _split_line($delimiter, $line);
75 foreach my $i ( 0 .. $#v ) {
76 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
78 push @{ $data->{items} }, $item;
81 $data->{header} = [ @header ];