A question about HTML::Parser.

[zhengmath]

I am sorry to bother again.
What I want to do is to input a htm file, then delete all the tables, strange html language symbol, and return a plain text.
What the code below did is:
It deletes all the html tags, like <TABLE>, <\Table>, <..> and etc, but it did not delete anything between <TABLE> and <\TABLE>, and there are some symbols like, &nbsp, $#151 and etc.
Please help me modify the file to do the job. Thanks very much!

       #!/usr/bin/perl -w
       use strict;
       use HTML::Parser;
       # define the subclass
       package IdentityParse;
       use base "HTML::Parser";
 
       my @processed_html;
 
       sub text {
          my ($self, $text) = @_;
          # print out the text
          push(@processed_html, $text);
      }
 
      sub comment {
      }
 
      sub start {
          my $self = shift;
          $self->{table_seen}++ if $_[0] eq  "table" ;
          $self->SUPER::start(@_);
 
      }
 
      sub end {
          my $self = shift;
          $self->SUPER::end(@_);
          $self->{table_seen}-- if $_[0] eq "table";
 
      }
      sub output
         {
         my $self = shift;
         unless ($self->{table_seen}) {
                   $self->SUPER::output(@_);
                }
         }
 
      my $p = new IdentityParse;
      $p->parse_file("H:/Test Data/wmt2004.htm");
      open OUT, '>', "H:/Test Data/text1-processed.txt" or die;
      print OUT @processed_html;
      close OUT;
 

[kleach]

@zhengmath
This will remove everything in tables.

while (<STDIN>) { $l.=$_; } $_=$l;
$_=$l;
while (/<table/is)
 { s/(.*)<table.*?>.*?<\/table>/$1/igs; }
 
print ;
 

The conversion of HTML::Entities to text is harder as many don't have text equiv. If you want to roll your own mapping many to a single character, just use s///.

[chorny]

IMHO, XPath is better suited for such job.