lib/CXGN/Tools/Parse.pm

   1 package CXGN::Tools::Parse;
   2 use strict;
   3
   4 =head1 CXGN::Tools::Parse
   5
   6  Interface-like base class for CXGN parsers.
   7
   8  Two different methods are encouraged:
   9  1) Give a filename to new(), and a filehandle is created,
  10     and you can parse one entry at a time w/ next().
  11  2) Send raw data to new(), and automatically parses everything
  12     at once with parse_all(), pushing each entry, as a hash ref,
  13         to @{$self->{entries}}
  14
  15 =head1 Author
  16
  17  C. Carpita <csc32@cornell.edu>
  18
  19 =head1 Methods
  20
  21 =cut
  22
  23 =head2 new()
  24
  25  Args: (opt) raw output data
  26  Ret: Parser object
  27  Side: Calls parse() automatically if argument provided
  28
  29 =cut
  30
  31 sub new {
  32         my $class = shift;
  33         my $self = bless {}, $class;
  34         my $data_or_file = shift;
  35         if(-f $data_or_file){
  36                 $self->{file} = $data_or_file;
  37                 my $fh;
  38                 open($fh, $self->{file}) or die "Can't open file for reading: " . $self->{file} . "\n";
  39                 $self->{fh} = $fh;
  40         }
  41         else{
  42                 $self->{data} = $data_or_file;
  43                 $self->{data_to_parse} = $data_or_file;
  44                 $self->parse_all_data();
  45         }
  46         return $self;
  47 }
  48
  49 sub parse_all {
  50         my $self = shift;
  51         while(my $entry = $self->next()){
  52                 push(@{$self->{entries}}, $entry);
  53                 $self->{entry_by_id}->{$entry->{id}} = $entry;
  54         }
  55 }
  56
  57 sub get_entry_by_id {
  58         my $self = shift;
  59         my $id = shift;
  60         return $self->{entry_by_id}->{$id};
  61 }
  62
  63 sub get_all_entries {
  64         my $self = shift;
  65         return @{$self->{entries}};
  66 }
  67
  68 sub next {
  69         my $self = shift;
  70
  71         #Do stuff with this:
  72         #my $data = $self->{data_to_parse};
  73         # grab entry, set hash, then...
  74         # $self->{data_to_parse} = $data_with_stuff_chopped_off
  75         # then return hashref
  76         #
  77         #or this, if filehandle exists:
  78         #
  79         #my $fh = $self->{fh};
  80         # do filehandle reads, grab entry, set hash
  81         # return hashref
  82         #
  83
  84         die "Override this function in a subclass";
  85
  86 }
  87
  88
  89 sub DESTROY {
  90         my $self = shift;
  91         $self->{fh}->close() if $self->{fh};
  92 }
  93
  94 1;