#!/usr/bin/perl

# baygrab.pl v0.2
# Josh Jackson <pturing@firehead.org>

#  Changelog
#  #########
#
#
#  2007.06.26 - 0.2
#    download either tags or albums
#    add cookie to allow downloading 'offensive' images
#
#  2007.06.22 - 0.1
#    downloading based on tags only
#    sort into album subdirectories
#


use strict;
use LWP::UserAgent;
use HTTP::Cookies;

my($mode) = 'tag';
my($album, $tag);

if ($ARGV[0] =~ /-album/) {
  $mode = 'album';
  $album = $ARGV[1];
  print "grabbing album $album\n";
}
elsif ($ARGV[0] =~ /-tag/) {
  $tag = $ARGV[1];
  print "grabbing tag $tag\n";
}
else {
  $tag = $ARGV[0];
  print "grabbing tag $tag\n";
}

my($filecounter);

sub init_agent {
  my($ua) = new LWP::UserAgent;
  $ua->agent("Mozilla/4.0 (compatible; IAmTheWalrus 5.5; OpenBDSM)");
  return $ua; 
}

sub request_cookies {
  my($request) = $_[0];
  my($cookie_jar) = HTTP::Cookies->new;
  my(@rest) = ("/", "bayimg.com", undef, 0, 0, 60*60, 0);
  $cookie_jar->set_cookie(undef, 'show_offensive', 'true', @rest);
  $cookie_jar->add_cookie_header($request);
  return $request;
}


sub get_to_file {
  my $url = $_[0];
  my $file_name = $_[1];
  my($ua) = init_agent();
  my($request) = request_cookies(HTTP::Request->new(GET => $url));
  my $res = $ua->request($request);
  open(FH, ">$file_name");
  print FH $res->content;;
  close(FH);
}


sub get_to_string {
  my($ua) = init_agent();
  my($url) = $_[0];
  my($request) = request_cookies(HTTP::Request->new(GET => $url));
  my $res = $ua->request($request);
  return $res->as_string;
}
  

sub post_to_string {
  my($ua) = init_agent();
  my $posturl = $_[0];
  my $postvars = $_[1];
  my $req = request_cookies(HTTP::Request->new(POST => $posturl));
  $req->content_type('application/x-www-form-urlencoded');
  $req->content($postvars);
  my $res = $ua->request($req);
  #print "requested address $posturl with vars $postvars\n";
  my($ret) = $res->as_string;
  #print "result is:\n$ret\n";
  return $ret;
}


sub fetchtag_page {
  my($tag) = $_[0];
  my($pagenum) = $_[1];
  if ($pagenum == '') { $pagenum = 1; }
  return post_to_string('http://bayimg.com/ajax_tags.php', "page=$pagenum&tag=$tag");
}


sub fetchalbum_page {
  my($album) = $_[0];
  my($pagenum) = $_[1];
  if ($pagenum == '') { $pagenum = 1; }
  return post_to_string('http://bayimg.com/ajax_album.php', "page=$pagenum&id=$album");
}


sub thumbpage_to_imagepages {
  my($x) = $_[0];
  my(@url_list) = ();
  my(@paths) = ($x =~ /href\=\"\/.........\"/g);
  foreach(@paths) {
    push(@url_list, substr($_, 7, 9));
  }
  return @url_list;
}

sub before_first {
  my(@parts) = split(/$_[0]/, $_[1]);
  return $parts[0];
}

sub after_first {
  my(@parts) = split(/$_[0]/, $_[1]);
  return $parts[1];
}

sub imagepage_imageurl {
  my($x) = $_[0];
  $x =~ /src\=\"(.*)\"\ id\=\"mainImage\"/;
  return $1;
}


sub imagepage_namefromtitle {
  my($x) = $_[0];
  $x =~ /\<title\>bayimg\ -\ image\:\ (.*)\ -\ free\ uncensored\ image\ hosting\<\/title\>/;
  #print "title found as $1\n";
  return $1;
}

sub imagepage_namefrominfo {
  my($x) = $_[0];
  $x = after_first('\<div\ id\=\"extra\"\>', $x);
  $x = before_first('\<h2\>Popular\ tags\ on\ the\ site\<\/h2\>', $x);
  $x =~ /\<p\>Image\:\ (.*)\<br\>/;
  #print "info line found as $1\n";
  return $1;
}


sub imagepage_imagename {
  my($id) = $_[1];
  my($x) = $_[0];
  my($fromtitle) = imagepage_namefromtitle($_[0]);
  my($frominfo) = imagepage_namefrominfo($_[0]);
  $filecounter += 1;
  my($c) = sprintf("%04d", $filecounter);

 
  if (length($fromtitle) == 0) {
    if (length($frominfo) == 0) {
      $x = $id;
    }
    else {
      $x = $frominfo;
    }
  }
  else {
    $x = $fromtitle;
  }

  if (length($frominfo) != 0 and length($fromtitle) != 0 and $frominfo != $fromtitle) {
    print "title ($fromtitle) differs from info ($frominfo)\n";
  }

  if ($x =~ /\.(rar|zip)$/) {
    mkdir($x);
    return "$x/$c-$id";
  }
  else {
    return "$c-$x";
  }
}


sub saveimage {
  my($page) = get_to_string("http://bayimg.com/$_[0]");
  my($image_name) = imagepage_imagename($page, $_[0]);
  print "Saving $image_name\n";
  get_to_file(imagepage_imageurl($page), $image_name);
}


sub last_page {
  my($x) = $_[0];
  $x =~ /.*parameters:\ \'page\=([0-9]+)\&.*parameters:\ \'page\=([0-9]+)\&/;
  print "last page is $1\n";
  return $1;
}


sub fetchpage_bymode {
  my($x) = $_[0];
  print "will fetch page $x in $mode mode\n";
  my($pagehtml) = '';
  if ($mode =~ /tag/) {
    $pagehtml = fetchtag_page($tag, $x);
  }
  elsif ($mode =~ /album/) {
     $pagehtml = fetchalbum_page($album, $x);
  }
  else {
    print 'error: mode not set\n';
  }
  return $pagehtml;
}



sub bayimg_fetchall {
  my(@image_pages);
  my($page) = 1;
  my($pagehtml) = fetchpage_bymode($page);
  my($lastpage) = last_page($pagehtml);
  for ($page = 1; $page <= $lastpage; $page++) {
    print "Page $page\n";
    my($pagehtml) = fetchpage_bymode($page);
    @image_pages = thumbpage_to_imagepages($pagehtml);
    my $count = @image_pages;
    print "$count images to grab\n";
    foreach(@image_pages) {
      saveimage($_);
    }
  }
}


bayimg_fetchall();