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 if ( $line =~ s/^"// ) {
23 $line =~ s/""/_qq_/gc;
24 $line =~ s/^\s*([^"]+)\s*"\Q$delimiter\E?// || die "can't parse $line";
26 } elsif ( $line =~ s/^\s*([^\Q$delimiter\E]+)\s*\Q$delimiter\E?// ) {
28 } elsif ( $line =~ s/^\s*\Q$delimiter\E// ) {
31 die "can't parse [$line]\n";
34 $v =~ s/^\s*(.+?)\s*$/$1/;
45 my $path = $self->full_path || $self->path;
47 my $data = read_file $path, { binmode => ':raw' }; # FIXME configurable!
48 my $encoding = 'utf-8';
49 if ( $path =~ m/\.(\w+).csv/i ) {
52 warn "decoding ", length($data), " bytes using $encoding\n";
53 $data = decode($encoding, $data);
55 my @lines = split(/\r?\n/, $data);
56 $data = { items => [] };
60 if ( $lines[0] !~ /;/ && $lines[1] =~ /;/ ) {
61 shift @lines; # FIXME ship non-header line
65 warn "$path ", $#lines + 1, " lines encoding: $encoding delimiter:",dump($delimiter);
67 my $header_line = shift @lines;
69 my @header = _split_line( $delimiter, $header_line );
70 warn "# header ",dump( @header );
72 while ( my $line = shift @lines ) {
74 my @v = _split_line($delimiter, $line);
76 foreach my $i ( 0 .. $#v ) {
77 $item->{ $header[$i] || "f_$i" } = [ $v[$i] ];
79 push @{ $data->{items} }, $item;
82 $data->{header} = [ @header ];