#!/usr/bin/perl -w
###########
# getcraig.pl
#
# A messy script to crawl craiglist for certain keywords and
# subsequently send notifications via e-mail of any matching
# finds.  Two levels are used: digest and notify.  Notify is
# for really hot stuff you definitely want to know about.
#
# I wrote this for keeping an eye out for deals on cars I'm
# interested in.
#
# Requires wget and mail
#
# Offered to public domain, since it's pretty silly.
#
# -JSK
# www.generalcriticism.com
# kutani at projectkutani com
###########

use strict;

# Enter your craigslist location here
my $location = "austin";

# Put the area of craigslist you want to seach here
my $section = "cta";


# Digest Addy is for somewhat-interested finds
my $digestaddy = "YOUR ADDRESS HERE";

# Notify Addy is for Hot Stuff
my $notifyaddy = "$digestaddy";


# Regexp for the digest search
my $digregexp = "is(| )(2|3)(0|5)0|300z|lotus|lancer evo|evolution|miata|porsche|alfa|romeo|240( |)sx";

# Regexp for the notify search
my $notregexp = "nsx| m3 |mr2|280z|RX(|-)7| 944|2\.5RS|mazdaspeed";




my $wget = "wget -q -O - ";
my $urlbase = "http://$location.craigslist.org/$section";
my $row  = 0;
my $ahref= 0; 
my $price= 0;
my $wrap = 0;
my $last = "NULL";
my $next = 1;
my $pg   = 0;

my @entries = ();
my @urls    = ();

if( -e "craig.last" ) {
	$last = `cat craig.last`;
	chomp($last);
}


my @page  = `$wget $urlbase/`;
TOP:
foreach my $line (@page) {
	if( ! $row && $line =~ /class="row"/ ) { $row = 1; }
	if( $row && $line =~ /<\/p>/ ) { $row = 0; $ahref = 0; next; }
		
	if( $wrap && $row ) {
		$line =~ s/^\s+//g;
		my @prs = split /<\/a/, $line;
		push @entries, $prs[0];
		$wrap = 0;
		next;
	}

	if( $row ) {
		if( ! $ahref && $line =~ /<a href="/ ) {
			$ahref = 1;
			my @prsa = split /">/, $line;
			my @prsb = split /"/, $prsa[0];
			@prsa = split /<\/a>/, $prsa[1];

			if( $prsb[1] eq $last ) { $next = 0; last; }
			if( $prsa[0] =~ /(\w|\d)+/ ) {
				push @entries, $prsa[0];
				push @urls,    $prsb[1];
			} else {
				$wrap = 1;
				push @urls,    $prsb[1];
			}
			next;
		}
	}
}

if( $next && $pg < 10000 ) {
	$pg += 100;
	#print STDERR "Getting $urlbase/index$pg.html\n";
	@page = `$wget $urlbase/index$pg.html`;
	goto TOP;
}

my @digest = ();
my @notify = ();

#print STDERR scalar @entries . " entries parsed.\n";
for( my $i = 0; $i < scalar @entries; $i++ ) {
	if( $i == 0 ) {
		open(FILE,">craig.last");
		print FILE "$urls[$i]";
		close(FILE);
	}	
	#print "$entries[$i] --- $urls[$i]\n";
	if( $entries[$i] =~ /$digregexp/i ) { push @digest, "$entries[$i] -- $urls[$i]"; }
	if( $entries[$i] =~ /$notregexp/i ) { push @notify, "$entries[$i] -- $urls[$i]"; }
}

if( scalar @digest ) {
	open(MAIL,"|mail -s \"Craigslist Digest\" $digestaddy");
	foreach( @digest ) { print MAIL "$_\n"; }
	close(MAIL);
}

if( scalar @notify ) {
	open(MAIL,"|mail -s \"Craigslist Hot Stuff\" $notifyaddy");
	foreach(@notify) { print MAIL "$_\n"; }
	close(MAIL);
}