#!/usr/bin/perl -w

use strict "vars";
use strict "refs";
use lib qw(/home/jenkster/perl); # Use local library

use HTTP::Request::Common qw(GET POST);
use HTTP::Cookies ();
use LWP::Simple ();
use LWP::UserAgent ();
use HTML::Form;

my ($request, $response, $html, $form, $record, $address);
my $row = 1;

my $userAgent = "Mozilla/5.0 (Windows; U; WinNT4.0; en-US; rv:1.4) Gecko/20030624 Netscape/7.1 (ax)";

my $ua = LWP::UserAgent->new;
my $cookie_jar = HTTP::Cookies->new(file => "cookies.txt", autosave => 1);

$ua->cookie_jar($cookie_jar);
$ua->agent($userAgent);

#$request = GET 'http://www.spb.ca.gov/employment/easy_search.htm';
#$request = GET 'http://www.spb.ca.gov/employment/search_p.cfm?showAll';

#$response = $ua->request($request);

#$cookie_jar->extract_cookies($response);

#open(OUT,">request.html");
#print OUT $response->content;
#close(OUT);

#$form = (HTML::Form->parse($response->content, 
#  'http://www.spb.ca.gov/employment/'))[1];

#$form->value('txtSearch','');
#$form->value('daterange','0');
#$form->value('order','title1');

#print $form->dump;

#$request = $form->click;

# $request->content_type('application/x-www-form-urlencoded');
# $request->header('Accept' => '*/*');
# $request->header('Allowed' => 'GET HEAD PUT');

#$response = $ua->request($request);

#$cookie_jar->extract_cookies($response);

#$html = $response->content;

#open(OUT,">search_result.html");
#print OUT $response->content;
#close(OUT);

#print "Length: ".length($html)."\n";

`mv *.html expired`;

do
{
  if ($row == 1)
  {
    $address = 'http://www.spb.ca.gov/employment/search_p.cfm?showAll';
  } else {
    $address = 'http://www.spb.ca.gov/employment/search_p.cfm?SHOWALL=&&sRow=' . $row . '&&searchType=http://spb.ca.gov/employment/wvpos_index.htm';
  }

  print "Address: $address\n";

  $request = GET $address;

  $response = $ua->request($request);

  $cookie_jar->extract_cookies($response);

  $html = $response->content;

  open(OUT,">request.html");
  print OUT $response->content;
  close(OUT);

  while ($html =~ m/more\_info\.cfm\?recno\=(\d+)/g)
  {
    if (-e "expired/$1.html")
    {
      print "Skipping Row: $row Record: $1\n";
      `mv expired/$1.html .`;
    } else {
      $request = GET "http://www.spb.ca.gov/employment/more_info.cfm?recno=$1";
      $response = $ua->request($request);
      $cookie_jar->extract_cookies($response);

      open(OUT,">$1.html");
      print OUT $response->content;
      close(OUT);

      print "Row: $row Record: $1 Content length:" . length($response->content) . "\n";
    }
    ++$row;
  }
}
while ($html =~ m/<input type="submit" value="   Next   ">/);

exit;
