#!/usr/bin/perl -w

use strict;
use LWP::UserAgent;

my $data = '/html/jobs/post-gazette-listings.html';
my %oldjobs = ();

open (JOBS, "<$data") || print "error opening $data\n";
  my $olddata = "";
if (open (JOBS, "+<$data"))
{
  read JOBS, $olddata, 500000;
  my @oj = ($olddata =~ /details.html\?id=([0-9a-f]*)/sg);
  print "number of old jobs found is ", scalar(@oj), "\n";
  map {$oldjobs{$_} = 'Y'} @oj;
}

my $ua = LWP::UserAgent->new;

my @newjobs = ();
my $next = "";

my $page;
do{
  my $urlbase = "http://pgh.careercast.com";
  my $url;
  if ($next)
  {
    $url = $urlbase . $next;
  }
  else
  {
    my $urlrt = 'js.php?lookid=pgh&qSort=date&qState=PA&view=2&pp=100';
  my $urlacc = 'qInd=pghcategoryaccounting';
  my $urladm = 'qInd=pghcategoryadministrative';
  my $urlcom = 'qInd=pghcategorycomputer';
    my $urledu = 'qInd=pghcategoryeducation';
  my $urlgen = 'qInd=pghcategorygeneral';
  my $urlpro = 'qInd=pghcategoryprofessional';
  my $urlhos = 'qInd=pghcategoryhospitality';
    $url = "$urlbase/$urlrt&$urlgen&$urlacc&$urladm&$urlcom&$urledu&$urlpro&$urlhos";
  }
  $| = 1;
    print " fetching page...";
  my $res = $ua->get($url);
    print "got it\n";
  $page = $res->content;
#  open NJ, "</tmp/general.html";
#    read NJ, $page, 500000;
    
  my @pageparts = split /<table[^>]*>/, $page;
  my $page2;
  foreach my $p (@pageparts)
  {
    if ($p =~ />Job Title</)
    {
      $page2 = $p;
      last;
    }
  }
  if (!defined($page2))
  {
    print "  Error: page2 not defined\n";
  }
  my @rows = split /<\/tr>/ , $page2;
    print "  Number of rows is ", scalar(@rows), "\n";
  foreach my $row (@rows)
  {
#    sleep 1;
#    my ($id, $desc, $date) =
#     ($row=~ /details.html\?id=([^&]*)[^>]*>([^<]*).*([0-9]*-[0-9]*-[0-9]*)/s);
#    print " row is |$row|\n";
    my ($id, $desc) =
      ($row=~ /details.html\?id=([^&]*)[^>]*>([^<]*)/s);
    if (!$id or !$desc)
    {
      next;
    }
#    print "  id is |$id| and desc is |$desc|\n";
    push @newjobs, [$id, $desc] unless (exists($oldjobs{$id}));
  }
  print "  found ", scalar(@newjobs), " new jobs so far\n";
} while (($next) = ($page =~ /<a href="([^>]*)"><img[^>]*but.arrowNext.gif/));

print ("number of new jobs found is ", scalar(@newjobs), "\n");

my (undef, undef, undef, $day, $mon, $year) = localtime;
$mon = (qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec))[$mon];
$year +=1900;
print "Date is $mon-$day-$year\n";

system "cp $data $data.bak";
open (JOBS, ">$data");
foreach my $job (@newjobs)
{
  print JOBS '<br><a href="http://pgh.careercast.com/texis/jobsearch/details.html?id=',
    $$job[0], '">', $$job[1], "</a> ";
  print JOBS "$mon-$day-$year\n";
}

print JOBS $olddata;