#!/usr/bin/perl
#    Last change:  PFB  15 Dec 99    0:53 am
#############################################################################
# impwww.cgi
# Imports Matt Wright's wwwboard messages into FutureForum Web Forum
# by Peter F. Brown - peterbrown@futurerealm.com
# FutureRealm Productions - futurerealm.com
# FutureForum Home Page: 'http://futurerealm.com/forum/futureforum.htm'
#############################################################################
# Copyright 1999 Peter F. Brown (FutureRealm Productions)
# Complies with the GNU GENERAL PUBLIC LICENSE
# and is released as "Open Source Software".
# You can therefore use it, modify it, and install it anywhere.
# You simply can't SELL IT directly, and you have to include the copyright.
# NO WARRANTY IS OFFERED FOR THE USE OF THIS SOFTWARE
#############################################################################

# NOTE! PLEASE... read the documentation below. This is a specialized script
#       that worked perfectly for me, but may not work for your implementation
#       of the wwwboard script. If you're not a Perl programmer, you may wish
#       to hire one to work through some of these issues.
#       -- This script is probably only worth messing with if you have a LOT
#       of wwwboard messages that you really, really want to keep.

# step 1: take the 'index.html' file (from wwwboard) and
#         remove all text above and below the outer <top> and <end>
#         tags. THIS IS MANDATORY. The script will assume that
#         all extraneous text is gone.

# ==>>>> THIS OF COURSE ASSUMES THAT YOU HAVE BACKED UP YOUR WWWBOARD FILES!

# step 2: set up your empty MySQL table prior to running this script
          # see the futureforum.cgi file for info about this

# step 3: edit all the variables below
# step 4: run the script

# FIELDS IN THE TABLE THAT IT WILL BE IMPORTED INTO:
# fields with no notes will be imported, also
#............................................................................
# message_no      # we'll use the message id numbers in the index.html file
# parent_no       
# branch_no       
# level           

# topic           # set this to the default of your choice
                  # since wwwboard won't have this field

# subject         
# author          
# email           # we have to pull this from the message html file

# date            
# time
# seconds         # we'll set this based on the above date and time

# body            # we have to pull this from the message html file
# responses       
# send_email      # we'll set this based on the default you choose
#................................................................................
# ===>>>> NOTE!! I've kept my own data in this file, that I used, when
#                I imported my wwwboard files for the "HeartThread Forum",
#                (at 'http://futurerealm.com/forum/heart_thread/'),
#                because I figured a real example would be the best way
#                to explain how to do it. So... just change things as you wish.
#................................................................................
# ====>>>> NOTE 2: This script is sparsely documented; however, it's all I
#                  could do at this point. Frankly, it was a real headache
#                  putting this script together... I'm glad I only had to
#                  use it once! :-) I never would have bothered, but I had
#                  over 500 messages that I didn't want to lose.
#................................................................................
# ====>>>> REALLY, REALLY IMPORTANT NOTE!!!!
# The logic of this script made a number of assumptions; one of the biggest
# was that the wwwboard index file and individual message files followed
# a specific format. One issue that WILL NEED ADJUSTING is the selection
# of the DELIMITERS around your body text in each individual message.
# In my copy of wwwboard, which I had heavily customized, the body text
# was surrounded by the <dl> characters, like this:

# <dl><dd><dl><dd>
# multi line body text here ....
# ...
# </dl></dl>

# This was NON-STANDARD, so you'll probably have to change it, and FIGURE
# OUT how to SELECT AND DELIMIT your body text. The code is toward the
# bottom of this script.

# IN ADDITION, there may be other issues; so... try it, and if it works, great!
# (It worked perfectly for my copy, but my copy was, after all, a customized
# version. Oy... why is life so complicated... :-)
#................................................................................

$indexfile  = "./index_long.html";
$ext        = ".htm";
$msg_path   = "./heart_thread/messages";
$send_email = 'y';
$time       = '12:00:00';
$topic      = 'Any Topic';

$db_main        = 'forum';              # name of database (not table)
$db_host        = 'localhost';          

$db_table       = 'heart_thread';       # each forum should have its own table  
$cnf_file = "../../.my.cnf";            # has user name and password
$user_group = 'heart_thread';           # part of .my.cnf

$branch_length  = 8;                    # keep this at 8
                                        # DON'T CHANGE WITHOUT CONSULTING DOCS!!!

use DBI;
&connect_to_db;

%months = (
          'January',    '01',
          'February',   '02',
          'March',      '03',
          'April',      '04',
          'May',        '05',
          'June',       '06',
          'July',       '07',
          'August',     '08',
          'September',  '09',
          'October',    '10',
          'November',   '11',
          'December',   '12'
          );

# end of variable set up
##############################################################################

$cls = `clear`;
print $cls;
print '-' x 75;
print "\n\n";

#............................................................

open (FILE, $indexfile) or die "can't open $indexfile";
@file = <FILE>;
close(FILE);

# => sample of data from a wwwboard index file:

# <!--top: 522--><li><img src="new_tiny.gif"><a href="messages/522.htm">10th anniverisary depression</a> - <b>Dian</b> <i>7/11/99</i>
# (<!--responses: 522-->0)
# <ul><!--insert: 522-->
# </ul><!--end: 522-->

$message_no = 0;
$parent    = 0;
$level     = 1;
$branch_no = 0;
$response_count = 0;
undef(@parent);
$parent[0] = 0;

$minus_branch_length = $branch_length - ($branch_length * 2);
$chop_branch         = ($branch_length + 1);
$minus_chop          = $chop_branch - ($chop_branch * 2);

# e.g: substr($parent_branch_no, -8, 8 ) = '';
# substr($parent_branch_no, $minus_chop, $chop_branch ) = '';


foreach $line (@file)
{
        if ($line =~ /<!--top: [\d]+/)
                {
                $message_no = $&;
                $message_no = substr($message_no,9);

                $subject = $line;
                $subject = substr($subject, (index($subject,'htm">') + 5));
                $subject = substr($subject, 0, index($subject,'</a>'));

                @branch_elements = split (/\-/, $branch_no);
                $branch_elements = @branch_elements;

                # print qq~
                # start loop:
                # level: $level
                # parent: $parent
                # br_no: $branch_no
                # branch elements: $branch_elements
                # \@branch elements: @branch_elements
                # ...................
                # ~;

                if ( $level == 1 )
                        {
                        $new_branch_no = $message_no;
                        }
                elsif ( $level == 2 and ( $branch_no !~ /\-/ ) )
                        {
                        $new_branch_no = 1;
                        }

                elsif ( $branch_elements < $level )
                        {
                        $new_branch_no = 1;
                        }

                elsif ( $level >= 2 and $branch_elements == $level )
                        {
                        # chop off only 1 element
                        $old_branch_base = $branch_no;
                        substr($old_branch_base, $minus_chop, $chop_branch ) = '';
                        $old_branch_incr = substr($branch_no, $minus_branch_length, ($branch_length));

                        $branch_no = $old_branch_base;
                        $new_branch_no = int($old_branch_incr) + 1;
                        }

                elsif ( $level >= 2 and $branch_elements > $level )
                        {
                        # chop off x elements

                        $chop_elements = $branch_elements - $level;
                        $chop_minus_big = $minus_chop * $chop_elements;
                        $chop_branch_big = $chop_branch * $chop_elements;

                        $old_branch_long = $branch_no;
                        substr($old_branch_long, $chop_minus_big, $chop_branch_big ) = '';

                        $old_branch_base = $old_branch_long;
                        substr($old_branch_base, $minus_chop, $chop_branch ) = '';

                        $branch_no = $old_branch_base;
                        $old_branch_incr = substr($old_branch_long, $minus_branch_length, ($branch_length));

                        $new_branch_no = int($old_branch_incr) + 1;
                        }
                 else
                        {
                        $new_branch_no = 1;
                        }

                #...................................................
                $lenbn = length("$new_branch_no");
                $lenbal = $branch_length - $lenbn;
                $pad = '0' x $lenbal;
                $new_branch_no = $pad . $new_branch_no;

                if ( $level == 1 )
                        {
                        $branch_no = $new_branch_no;
                        }
                else
                        {
                        $branch_no = "$branch_no\-$new_branch_no";
                        }

                # print qq~
                # end of loop:
                # message_no: [$message_no]
                # branch_no: [$branch_no]
                # old br base: $old_branch_base
                # old br incr: $old_branch_incr
                # level: $level
                # parent: $parent[$level-1]

                }
        elsif ($line =~ /<!--responses: [\d]+\-->[\d]+/)
                {
                $responses = $&;
                $responses = substr($responses, index($responses, '-->')+3 );

                $parent_no = $parent[$level-1];

                print '-' x 75;
                print "\n";
                print "P: $parent_no - L: $level - M: $message_no - R: $responses - B: $branch_no\n";
                print "............. [$subject] ......\n";

                # insert into db
                &insert_msg;

                }
        elsif ($line =~ /<!--insert: [\d]+/)
                {
                $insert = $&;
                $insert = substr($insert, 12 );
                $level++;
                $parent[$level-1] = $message_no;
                $response_count++;
                }
        elsif ($line =~ /<!--end: [\d]+/)
                {
                $end = $&;
                $end = substr($end, 9 );
                $level--;
                $response_count--;
                if ( $level == 1 )
                        {
                        $parent = 0;
                        undef(@parent);
                        $parent[0] = 0;
                        }

                }

        else
                {
                print "\n Error!!! \n";
                }

}

print "\n\n";

# now fix responses
#...........................................................................

$sth_s = $dbh->prepare("select message_no, level, branch_no
                        from $db_table
                        order by message_no
                        ")
                        or
                        die "Can't do select: $DBI::errstr\n";

$rv = $sth_s->execute or die "Can't execute the query: $DBI::errstr\n";

while (( $message_no, $level, $branch_no ) = $sth_s->fetchrow_array )
        {

        $sth_s2 = $dbh->prepare("select count(*)
                                 from $db_table
                                 where
                                 branch_no like '$branch_no%'
                                 and
                                 level = ($level + 1)
                                ")
                                or die "Can't do select: $DBI::errstr\n";

        $rv2 = $sth_s2->execute or die "Can't execute the query: $DBI::errstr\n";
        ( $responses ) = $sth_s2->fetchrow_array;

        $sth_s3 = $dbh->do("update $db_table
                            set responses = $responses
                            where
                            message_no = $message_no
                            ") or
                            die "Can't save data: $DBI::errstr\n";
    
        warn $DBI::errstr if $DBI::err;

        }

#..........................

exit;

##################################################################
# connect_to_db
sub connect_to_db
{

$dbh = DBI->connect("DBI:mysql:$db_main:$db_host"
                     . ";mysql_read_default_file=$cnf_file"
                     . ";mysql_read_default_group=$user_group",
                     $user, $password)
                     || die print qq~
                     \n
                     \n
                     Error connecting to the Forum.
                     \n
                     ~;

warn $DBI::errstr if $DBI::err;

}
##################################################################
# insert_msg
sub insert_msg
{

# responses,

if ( $subject =~ /step|ex/i ){$topic = 'Step Families';}
elsif ( $subject =~ /marriage|wedding/i ){$topic = 'Marriage Issues';}
elsif ( $subject =~ /statistic/i ){$topic = 'Family Statistics';}
elsif ( $subject =~ /teen/i ){$topic = 'Teenagers';}
elsif ( $subject =~ /baby|babies/i ){$topic = 'Babies';}
elsif ( $subject =~ /parent/i and $subject !~ /step/i ){$topic = 'Parenting Issues';}
elsif ( $subject =~ /culture/i ){$topic = 'The Culture of Heart';}
elsif ( $subject =~ /school/i ){$topic = 'General Education';}
else {$topic = 'Any Topic';}

$author         = '';
$email          = '';
$date           = '';
$time           = '';
$date_time      = '';
$body           = '';
$new_date_time  = '';
$fixed_date     = '';

local $/;

open (MESSAGE, "$msg_path/$message_no$ext") or die "Can't open $message_no\n";
$message = <MESSAGE>;
close(MESSAGE);

# <input type=hidden name="origname" value="Sherry">
# <input type=hidden name="origemail" value="shadows32@sprintca.com">
# <input type=hidden name="origdate" value="March 12, 1999 at 19:12:56">

$message =~ /<input type=hidden name="origname" value="[^>]*>/;
$author = $&;
$author = substr($author,42);
substr($author,-2,2) = '';

$message =~ /<input type=hidden name="origemail" value="[^>]*>/;
$email = $&;
$email = substr($email,43);
substr($email,-2,2) = '';

$message =~ /<input type=hidden name="origdate" value="[^>]*>/;
$date_time = $&;
$date_time = substr($date_time,42);
substr($date_time,-2,2) = '';

# <input type=hidden name="origdate" value="March 12, 1999 at 19:12:56">
($date, $time) = split (/ at /, $date_time, 2);
($month, $day, $year) = split (/ /, $date, 3);

$day =~ s/,//;
if ( length($day) == 1 ){$day = '0' . $day;}

$month = $months{$month};
$fixed_date = "$year-$month-$day";

$new_date_time = "$fixed_date $time";

# ==> If this doesn't reflect the setup of your messages, you'll
#     have to change it, or it won't work!!!

# <dl><dd><dl><dd>
# body
# </dl></dl>

$body = substr($message, (index($message,'<dl><dd><dl><dd>') + 16));
$body = substr($body,0, index($body,'</dl></dl>') );

# print qq~
# author: $author
# email:  $email
# date_time: $date_time
# date: $date
# fixed_date: $fixed_date
# time: $time
# new_date_time: $new_date_time
# body:
# $body
# ~;

$field_str = "message_no,parent_no,branch_no,level,topic,subject,send_email,
              author,email,date,time,seconds,body
             ";

$subject = quotemeta($subject);
$author  = quotemeta($author);
$email   = quotemeta($email);
$body    = quotemeta($body);

$value_str = "'$message_no',
              '$parent_no',
              '$branch_no',
              '$level',
              '$topic',
              '$subject',
              '$send_email',
              '$author',
              '$email',
              '$fixed_date',
              '$time',
              unix_timestamp('$new_date_time'),
              '$body'
             ";
    
$sth_s = $dbh->do("insert into $db_table ($field_str) values ($value_str)") or
               die "Can't save data: $DBI::errstr\n";
    
warn $DBI::errstr if $DBI::err;

}
##################################################################