html-dom.pl to HTML.

index -|- end

Generated: Sun Aug 21 11:11:06 2011 from html-dom.pl 2011/01/12 3.7 KB.

#!/usr/bin/perl -w
# NAME: html-dom.pl
# AIM: Test HTML::DOM interface...
use strict;
use warnings;
use File::Basename;  # split path ($name,$dir,$ext) = fileparse($file [, qr/\.[^.]*/] )
use Cwd;
use HTML::DOM;
use Data::Dumper; # prt( Dumper($root) );
my $perl_dir = 'C:\GTools\perl';
unshift(@INC, $perl_dir);
require 'logfile.pl' or die "Unable to load logfile.pl ...\n";
# log file stuff
my ($LF);
my $pgmname = $0;
if ($pgmname =~ /(\\|\/)/) {
    my @tmpsp = split(/(\\|\/)/,$pgmname);
    $pgmname = $tmpsp[-1];
}
my $outfile = $perl_dir."\\temp.$pgmname.txt";
open_log($outfile);

# user variables
my $debug_on = 0;
my $def_file = 'c:\GTools\perl\html-dom.htm';
my $load_log = 1;
my $in_file = $def_file;


### program variables
my @warnings = ();
my $cwd = cwd();
my $os = $^O;

sub show_warnings($) {
    my ($val) = @_;
    if (@warnings) {
        prt( "\nGot ".scalar @warnings." WARNINGS...\n" );
        foreach my $itm (@warnings) {
           prt("$itm\n");
        }
        prt("\n");
    } else {
        prt( "\nNo warnings issued.\n\n" );
    }
}

sub pgm_exit($$) {
    my ($val,$msg) = @_;
    if (length($msg)) {
        $msg .= "\n" if (!($msg =~ /\n$/));
        prt($msg);
    }
    show_warnings($val);
    close_log($outfile,$load_log);
    exit($val);
}


sub prtw($) {
   my ($tx) = shift;
   $tx =~ s/\n$//;
   prt("$tx\n");
   push(@warnings,$tx);
}

sub get_html_dom($) {
    my ($inf) = @_;
    my $dom_tree = new HTML::DOM;
    prt("Parsing file [$inf]...\n");
    $dom_tree->parse_file($inf);
    prt("Done file [$inf]...\n");
    #prt( $dom_tree->innerHTML . "\n" );
    #prt( Dumper($dom_tree) . "\n" );
    my $name = "*";
    my $tree = $dom_tree->getElementsByTagName ( $name );
    prt( Dumper($tree) . "\n" );
}


sub process_in_file($) {
    my ($inf) = @_;
    if (! open INF, "<$inf") {
        pgm_exit(1,"ERROR: Unable to open file [$inf]\n"); 
    }
    my @lines = <INF>;
    close INF;
    my $lncnt = scalar @lines;
    prt("Processing $lncnt lines, from [$inf]...\n");
    my ($line,$inc,$lnn);
    $lnn = 0;
    foreach $line (@lines) {
        chomp $line;
        $lnn++;
    }
    get_html_dom($inf);
}

#########################################
### MAIN ###
# parse_args(@ARGV);
# prt( "$pgmname: in [$cwd]: Hello, World...\n" );
process_in_file($in_file);
pgm_exit(0,"Normal exit(0)");
########################################
sub give_help {
    prt("$pgmname: version 0.0.1 2010-09-11\n");
    prt("Usage: $pgmname [options] in-file\n");
    prt("Options:\n");
    prt(" --help (-h or -?) = This help, and exit 0.\n");
}
sub need_arg {
    my ($arg,@av) = @_;
    pgm_exit(1,"ERROR: [$arg] must have following argument!\n") if (!@av);
}

sub parse_args {
    my (@av) = @_;
    my ($arg,$sarg);
    while (@av) {
        $arg = $av[0];
        if ($arg =~ /^-/) {
            $sarg = substr($arg,1);
            $sarg = substr($sarg,1) while ($sarg =~ /^-/);
            if (($sarg =~ /^h/i)||($sarg eq '?')) {
                give_help();
                pgm_exit(0,"Help exit(0)");
            } else {
                pgm_exit(1,"ERROR: Invalid argument [$arg]! Try -?\n");
            }
        } else {
            $in_file = $arg;
            prt("Set input to [$in_file]\n");
        }
        shift @av;
    }

    if ((length($in_file) ==  0) && $debug_on) {
        $in_file = $def_file;
    }
    if (length($in_file) ==  0) {
        pgm_exit(1,"ERROR: No input files found in command!\n");
    }
    if (! -f $in_file) {
        pgm_exit(1,"ERROR: Unable to find in file [$in_file]! Check name, location...\n");
    }
}

# eof - template.pl

index -|- top

checked by tidy  Valid HTML 4.01 Transitional