#!/usr/bin/perl

use Mysql;
use Getopt::Long;  # http://perl.active-venture.com/lib/Getopt/Long.html

##################################################
#     EXPORT WORDPRESS FROM A MOVABLE TYPE DATABASE WITHOUT A WEB SERVER
#
#  by Brian Doom - Filmmaker, programmer, candy entrepreneur, business maniac.
#  brain @ ikillspies.com
#  comment on my blog if you need help, 
#    but for the most part this script is unsupported
#
##################################################

sub usage () {
  print "Generates a file from an ancient Movable Type database\n";
  print "Completely unsupported.\n";
  print "Hardcoded category, blog_id, database login info, timezone\n";
  print "Imports comments but not trackbacks.\n";
  print "Sets all entries to a single category.\n\n";
  print "To use:  \n";
  print "\t0) read entire script!  Set your default category.\n";
  print "\t1) Edit the file and set all the database login info.  Set Blog Id.\n";
  print "\t2) run!\n";
  print "\t3) your file is still not done.  Make a XML export of your current WordPress blog, this will be the 'real' file.\n";
  print "\t4) from the 'real' file, copy the header down to but not including the first <item> tag\n";
  print "\t5) prepend to your generated file\n";
  print "\t6) add '</channel></rss>' to the end of your file \n";
  print "\t7) upload to WordPress\n";
  print "\t\t7a) may time out.  In this case, go to blog admin and wait for number of entries to stop increasing\n";
  print "\t\t7b) eliminate successful entries from file and repeat submission!\n";
  print "\n";
} # end sub usage

# read in which blog we are doing
my $blogId = '';
my $help = '';
my $outFile = 'custom-MT-Export.xml';
my $logFile = 'customMTExport.log';

GetOptions ('help|?' => \$help);
if ($help) {
  usage();
  exit(0);
}

GetOptions ('blog|id|b=i' => \$blogId);
GetOptions ('out' => \$outFile);
GetOptions ('log' => \$logFile);

open ( OUT, ">$outFile" ) or die "ERROR: couldn't open $outFile for output\n";
open ( LOG, ">>$logFile" ) or die "ERROR: couldn't open $logFile for logging\n";

# open DB connection
# REMEMBER TO SET THESE BEFORE RUNNING!!!
$host = 'localhost';
$database = 'blog';
$user = 'mtblogtest';
$password = 'fnord';

# use either MySQL or DBI... not both!

# MySQL:
# $connection = Mysql->connect($host, $database, $user, $password) 
#  or die "could not connect to $database using $user@$host \n";

# DBI:
$dsn = "dbi:mysql:$database:localhost:3306";
$connection = DBI->connect ($dsn, $user, $password)
		or die "could not connect to $database using $user@$host \n";

# see the "4" here?  Set that to your blog key in mt_blog.blog_id
# GMT is set to California time because we are the center of the world
$query = <<ENDQUERY;
SELECT e.entry_id, e.entry_status,
       authors.author_nickname, 
       e.entry_title, e.entry_excerpt, e.entry_text, 
       e.entry_text_more, e.entry_created_on,
		   DATE_ADD(e.entry_created_on, INTERVAL 8 HOUR)
FROM blog.mt_entry e, 
     blog.mt_author authors
WHERE e.entry_blog_id=4
  AND e.entry_author_id = authors.author_id
ORDER BY entry_id
ENDQUERY



$statement = $connection->prepare($query)
  or die "Preparing SQL failed!  Query was $query \n";

$statement->execute()
  or die "Executing SQL failed!  Query was $query \n";

$statement->bind_columns( \$entry_id, \$entry_status,
       \$entry_author, 
       \$entry_title, \$entry_excerpt, \$entry_text, 
       \$entry_text_more, \$entry_created_on,
													\$entry_created_on_gmt);


# header: WordPress
# this is actually quite extensive: maybe put it in manually?
#print OUT '<?xml version="1.0" encoding="UTF-8"?>' . "\n";


# DEBUG
#$i = 0;

while ( $entry = $statement->fetchrow_hashref() ) {

  # DEBUG: reduce number of iterations for test
		#$i++;
  #last if ($i > 30);

  print OUT "<item>\n";

  # TODO: for all fields, do not output if empty

  # MISSING FIELD: link

  print OUT "<title>$entry_title</title>\n";
  print OUT "<pubDate>$entry_created_on</pubDate>\n";
  print OUT "<dc:creator>$entry_author</dc:creator>\n";

  # NOTE: for some reason, we lost all category data in the MT database

  # default: make them all "looky"
  print OUT "<category><![CDATA[looky]]></category>\n";


  # NOTE: if you want to do categories, now would be a good time

  # MISSING FIELD: guid

  print OUT "<description></description>\n<content:encoded><![CDATA[";


	# don't print anything if it's blank!

	if ( ( $entry_text !~ /^\s*$/ ) || ( $entry_text_more !~ /^\s*$/ ) ) {
			print OUT "$entry_text \n  $entry_text_more";
	} else {
			#print OUT "DEBUG: length zero!  $entry_text $entry_text_more\n";
	}

  # handle "link-only" posts:
  if ( $entry_excerpt =~ /^\w*http.*/ ) {
    print OUT "Web link of note: ";
    print OUT "<A HREF='$entry_excerpt'>$entry_title</A> \n";
    print OUT "(At $entry_excerpt)<P />\n"
	} elsif ( $entry_excerpt =~ /^\s*$/ ) {
			# don't print anything if it's blank!
			#print OUT "DEBUG: length zero!  $entry_excerpt \n";
	} else {
			print OUT "\n\n<I>$entry_excerpt</I>\n";
	} # end "link-only"

  print OUT "]]></content:encoded>\n";

# MISSING:
#<wp:post_id>6</wp:post_id>
#<wp:post_date>2007-10-24 19:25:25</wp:post_date>
#<wp:post_date_gmt>2007-10-25 02:25:25</wp:post_date_gmt>
#<wp:post_name>chicago-day-4-living-maxim-and-geekery</wp:post_name>

print OUT <<BLOCK;
<wp:post_date>$entry_created_on</wp:post_date>
<wp:post_date_gmt>$entry_created_on_gmt</wp:post_date_gmt>
<wp:comment_status>open</wp:comment_status>
<wp:ping_status>open</wp:ping_status>
<wp:status>publish</wp:status>
<wp:post_parent>0</wp:post_parent>
<wp:menu_order>0</wp:menu_order>
<wp:post_type>post</wp:post_type>
BLOCK


  # fetch entry comments

  # TODO: SQL $commentDateGMT
  $commentquery = <<COMMENT2;
    SELECT c.comment_author, c.comment_created_on, 
	         DATE_ADD(c.comment_created_on, INTERVAL 8 HOUR),
           c.comment_ip, c.comment_email,
           c.comment_url, c.comment_text
    FROM mt_comment c
    WHERE c.comment_entry_id = $entry_id
COMMENT2

  $statement2 = $connection->prepare($commentquery)
    or die "Preparing SQL failed!  Query was $query \n";

  $statement2->execute()
    or die "Executing SQL failed!  Query was $query \n";

  $statement2->bind_columns( \$commentAuthor, 
														 \$commentDate, \$commentDateGMT,
														 \$commentIP, \$commentEmail, 
														 \$commentUrl, \$commentBody );

  while ( $comment = $statement2->fetchrow_hashref() ) {
    print OUT <<BLOCK2;
<wp:comment>
<wp:comment_id>1</wp:comment_id>
<wp:comment_author><![CDATA[$commentAuthor]]></wp:comment_author>
<wp:comment_author_email>$commentEmail</wp:comment_author_email>
<wp:comment_author_url>$commentUrl</wp:comment_author_url>
<wp:comment_author_IP>$commentIP</wp:comment_author_IP>
<wp:comment_date>$commentDate</wp:comment_date>
<wp:comment_date_gmt>$commentDateGMT</wp:comment_date_gmt>
<wp:comment_content>$commentBody</wp:comment_content>
<wp:comment_approved>1</wp:comment_approved>
<wp:comment_type></wp:comment_type>
<wp:comment_parent>0</wp:comment_parent>
</wp:comment>
BLOCK2

  } # end for every comment for this entry

	print OUT "</item>\n";

	$statement2->finish;

  # that is the end of all the data for a single entry

  # progress meter
  print LOG ".";

} # end for every entry

$statement->finish;

$connection->disconnect();

close LOG;
close OUT;