Application Source Code

Analyse Page /cgi-bin/analpage.pl

Return to the notes... - Run the program - View dwd.pm module code - View dwd.conf external config file
#!/usr/local/bin/perl -w

#

#  analpage.pl    (c) Andy Belcher DandyWebDesign 2006

#

#  Building on another application; getpage.pl, this will return the

#  source code of the requested URL and parse the HTML into an

#  accessible data-structure.

#

######################################################################

use strict;

use CGI qw(:all);

use LWP::UserAgent;

use dwd;



#  INITIALISE PROGRAM ENVIRONMENT

my %G;      #  Stores global values within the program.

my %arg;    #  Stores the CGI form/URL parameters

my @parameters  = qw(requrl);

initprogenv(\%G,\%arg,\@parameters);



#  DEFINE PROGRAM SPECIFIC VALUES

$G{'pagetitle'}    = "Analyse Page";

$G{'title'}        = "$G{'title'} $G{'pagetitle'}";

$G{'returnurl'}    = "/tech/web/webanalpage.shtml";

$G{'sourceurl'}    = "/tech/web/analpage_source.html";

$G{'pmsourceurl'}  = "/tech/web/dwdpm_source.html";

$G{'requrl'}       = "";

my $error          = "";

my @html           = qw();

#  GET PAGE

if($arg{'requrl'})

  {$G{'requrl'} = $arg{'requrl'};

  if(useragent_fetchpage(\%G)) {parsehtml(\@html,$G{'uacontent'});}

  else {$error = "<p class=\"loud\"><strong>ERROR!</strong> $G{'uaerror'}</p>";}

  }

else {$arg{'requrl'} = "http://dwd/tech/web/web_index.shtml";}



  OUTPUT PAGE

&do_topsection;

print<<HERE;

<div class="main">

<p>Enter a valid URL below...</p>

<p><form method="post" action="$G{'progname'}">

<input type="text" name="requrl" size="50" value="$arg{'requrl'}" />

<br /><input type="submit" value="Get the page..." class="button"

  /></form></p>

</div><!--main-->

$error

HERE

  my $ctr = 0;

  if(@html)

    {print "<table style=\"font-size:8pt;font-family:courier,lucida console,verdana,georgia,arial,Comic Sans MS,courier,lucida console,sans serif;\">";

    foreach my $el (@html)

      {print "<tr><td>$ctr</td>";

      if(($ctr/2 == int($ctr/2)) || ($el =~ /^\s*<\s*!DOCTYPE/gsix))

        {$el = dehtml_string($el);

        $el =~ s/\n/<strong>\\n<\/strong>\n/gsix;

        print "<td>&nbsp;</td><td><code>".$el."</code></td></tr>\n";

        }

      else

        {if($el =~ m/^<\s*\//)

          {print "<td><strong>".dehtml_string($el)."</strong></td><td>&nbsp;</td></tr>\n";}

        elsif($el =~ m/^<!--/)

          {print "<td>&nbsp;</td><td>".dehtml_string($el)."</td></tr>\n";}

        else

          {my %attrib;parseattrib(\%attrib,$el);

          print "<td><strong>$attrib{'element'}</strong></td><td>&nbsp;</td></tr>\n";

          foreach my $at (sort keys %attrib)

            {if($at eq "element") {next;}

            print "<tr><td>&nbsp;</td><td>$at</td><td>$attrib{$at}</td></tr>\n";

            }

          }

        }

      $ctr++;

      }

    print "</table>\n";

    }

  &do_bottom;

#  END PROGRAM

#

#  SUB-ROUTINES

######################################################################

sub do_top {do_std_head(\%G);print "</head>\n<body>\n";do_std_header(\%G);}



######################################################################

sub do_bottom {do_std_bottom($G{'footerssi'});exit;}



######################################################################

sub do_topsection

  {&do_top;print<<HERE;

<div id="pathbar"><script type="text/javascript">drawPathBar(

  'tech','/tech/tech.shtml','To the Technical pages index...',

  'web applications','/tech/web/web_index.shtml','To the web applications index...',

  'program notes','$G{'returnurl'}','The accompanying notes for this program...',

  'analpage.pl');//</script></div><!--pathbar-->

<div id="content">

<div id="rightbar"><dl class="nbar">

<dt>Application Links</dt>

  <dd><a href="$G{'returnurl'}"

    title="Return to the documentation for this application..."

    >Program Notes</a></dd>

  <dd><a href="$G{'sourceurl'}"

    title="View the source code for this application..."

    >source code</a></dd>

  <dd><a href="$G{'pmsourceurl'}"

    title="View the perl module source code..."

    >dwd.pm source</a></dd>

</dl></div><!--rightbar-->

<h2>$G{'pagetitle'}</h2>

HERE

  }



######################################################################

#  EOF