#!/usr/bin/perl

#
# Upgrade version 1 CNID databases to version 2
#
# $Id: cnid2_create.in,v 1.2 2005-04-28 20:49:19 bfernhomberg Exp $
#
# Copyright (C) Joerg Lenneis 2003
# All Rights Reserved.  See COPYING.
#
#

use strict;


### Globals

my $toplevel = $ARGV[0];

# Assume our current directory is .AppleDB in the share directory as the default. 

$toplevel = ".." unless $toplevel;

# Main file information data structure. Each entry in did_table points
# to a list of hashes. Key is a DID and each list member is a hash
# describing a file or directory in the directory corresponding to the
# DID. n_entries is the number of items found, output_list contains
# all entries that are eventually written to the output file.

my %did_table;
my $n_entries;
my @output_list;

# RootInfo values in the new format

my $ri_cnid    = "00000000";
my $ri_dev     = "1122334400000000";
my $ri_ino     = "0000000000000000";
my $ri_did     = "00000000";
my $ri_hexname = "526f6f74496e666f00";  # RootInfo\0


# Current CNID found in didname.db
my $current_cnid;

# Largest CNID from cnid.db, used if we cannot find the current CNID in didname.db 
my $max_cnid;

# Number of bogus/invalid entries found in the DB 
my $errors_found;

# Filenames

my $didname_dump = "didname.dump";
my $cnid_dump    = "cnid.dump";
my $cnid2_dump   = "cnid2.dump";

### Subroutines

sub skip_header($)
{
    my $handle = shift;
    my $header_ok;

    while (<$handle>) {
	chomp;
	if ($_ eq "HEADER=END") {
	    $header_ok = 1;
	    last;
	}
    }
    die "No valid header found, invalid input file?" unless $header_ok;
}

sub print_eentry($$)
{
    my $reason = shift;
    my $entry  = shift;

    printf "??:%s %-9s%-9s%-9s%-9s%s\n",
    $reason,
    $entry->{cnid},
    $entry->{dev},
    $entry->{ino},
    $entry->{did},
    $entry->{name};
    
    $errors_found++;
}

sub find_current_cnid()
{
    my $key;
    my $val;
    my $cnid;
    my $compare = " " . $ri_did . $ri_hexname;

    # get rid of "00" at the end, RootInfo in the old format does not have a trailing \0

    $compare =~ s/00$//;
    
    open DIDNAME, "< $didname_dump" or die "Unable to open $didname_dump: $!";
    skip_header(*DIDNAME);

    while (1) {
	$key = <DIDNAME>;
	chomp $key;
	last if $key eq "DATA=END";
	$val = <DIDNAME>;
	chomp $val;
	last unless defined($key) and defined($val);	
	if ($key eq $compare) {
	    # \00\00\00\00RootInfo
	    $val =~ s/^ //;
	    $cnid = $val;
	    last;
	}	
    }
    close DIDNAME;
    return $cnid;
}

sub verify_entry($)
{
    my $entry = shift;

    if (length($entry->{cnid}) != 8 or length($entry->{name}) == 0) {
	print_eentry("fmt", $entry);
	return 0;
    } else {
	return 1;
    }
}

sub create_did_table()
{    
    my $data_ok;
    my $key;
    my $val;
    my $len;    
    my $i;
    my $name;
    my $cmax;

    open CNID, "< $cnid_dump" or die "Unable to open $cnid_dump: $!";
    skip_header(*CNID);

    while (1) {
	$key = <CNID>;
	chomp $key;
	$key =~ s/^ //;
	if ($key eq "DATA=END") {
	    $data_ok = 1;
	    last;
	}
	my $val = <CNID>;
	chomp $val;
	$val =~ s/^ //;

	last unless defined($key) and defined($val);
    
	# We do not worry about converting any of the
	# integer values into binary form. They are in network byte order, 
	# so we know how to extend them. We just treat them as hexadecimal ASCII strings.
	# The file name is also stored as a proper string, since we need to 
	# look for it in the file system.

	my $entry;
    
	$entry->{cnid}    = $key;
	$entry->{dev}     = substr($val, 0,   8);
	$entry->{ino}     = substr($val, 8,   8);
	$entry->{did}     = substr($val, 16,  8);
	$entry->{hexname} = substr($val, 24);
    
	$len = length($entry->{hexname}) - 2;
	$i = 0;
	$name = '';
	while ($i < $len) {
	    $name .= chr(hex(substr($entry->{hexname}, $i, 2)));
	    $i += 2;
	}
	$entry->{name} = $name;

	if (verify_entry($entry)) {
	    push @{$did_table{$entry->{did}}}, $entry;
	    $n_entries++;
	    $cmax = $entry->{cnid} if $entry->{cnid} gt $cmax;
	}
    }
    close CNID;
    die "No valid end of data found, invalid input file?" unless $data_ok;
    return $cmax;
}

sub output_header($$$)
{
    my $handle = shift;
    my $dbname = shift;
    my $n      = shift;
    
    printf $handle "VERSION=3\nformat=bytevalue\ndatabase=%s\ntype=btree\nHEADER=END\n", $dbname;
}

sub expand_did_table($$)
{
    my $did = shift;
    my $basename = shift;
    my $entry;
    my $name;

    foreach $entry (@{$did_table{$did}}) {
	$name = $basename . "/" . $entry->{name};

	if ( -e $name ) { 
	    if ( -d $name ) { 
		$entry->{type} = "00000001"; 
	    } else { 
		$entry->{type} = "00000000"; 
	    } 
	} else { 
	    
            # The file/dir does not exist in the file system. This could result
	    # from a non-afpd rename that has not yet been picked up by afpd and is
	    # fixable. We need a guess at the type, though.
	    
	    if ($did_table{$entry->{cnid}} and scalar(@{$did_table{$entry->{cnid}}}) > 0) {

		# We have entries hanging off this entry in our table,
		# so this must be a directory
		$entry->{type} = "00000001";
	    } else {
		# If this is actually an empty directory that was renamed, 
		# the entry will be deleted by afpd on the next access, 
		# so this should be safe 
		$entry->{type} = "00000000";		
	    }
	}

	$entry->{reached} = 1;
	push @output_list, $entry;
	expand_did_table($entry->{cnid}, $name);
    }
}

sub find_unreachable()
{
    my $did;
    my $list;
    my $entry;

    while (($did, $list) = each %did_table) {
	foreach $entry (@{$list}) {
	    print_eentry("reach", $entry) unless $entry->{reached};
	}
    }
}

sub find_duplicates()
{
    my %seen_cnid;
    my %seen_devino;
    my %seen_didname;    
    my $cnid;
    my $devino;
    my $didname;
    my $err;
    my $entry;
    
    for (my $i = 0; $i < scalar(@output_list); $i++) {
	$err = 0;
	$entry = $output_list[$i];
	$cnid = $entry->{cnid};
	$devino = $entry->{dev} . $entry->{ino};
	$didname = $entry->{did} . $entry->{name};
	if ($seen_cnid{$cnid}) {
	    print_eentry("exist_cnid", $entry);
	    $err++;
	}
	if ($seen_cnid{$cnid}) {
	    print_eentry("dupl_cnid", $entry);
	    $err++;
	}
	if ($seen_devino{$devino}) {
	    print_eentry("dupl_devino", $entry);
	    $err++;
	}
	if ($seen_didname{$didname}) {
	    print_eentry("dupl_didname", $entry);
	    $err++;
	}
	if ($err) {
	    splice(@output_list, $i, 1);
	} else {
	    $seen_cnid{$cnid} = 1;
	    $seen_devino{$devino} = 1;
	    $seen_didname{$didname} = 1;
	}
    }
}


print "Finding the current CNID from $didname_dump\n";
$current_cnid = find_current_cnid();
print "Reading in entries from $cnid_dump\n";
$max_cnid = create_did_table();
print "Found $n_entries entries\n";

if (not $current_cnid) {
    print "Warning: could not find a valid entry for the current CNID in $didname_dump.\n";
    print "Using maximum CNID value $max_cnid from $cnid_dump instead\n";
    $current_cnid = $max_cnid;
} else {
    print "Current CNID is $current_cnid\n";
}

print "Building directory tree according to cnid.db\n";
expand_did_table("00000002", $toplevel);

if ($n_entries > scalar(@output_list)) {
    print "Looking for unreachable nodes in the directory tree:\n";
    find_unreachable();
}

print "Looking for duplicate entries\n";
find_duplicates();

print "Creating $cnid2_dump\n";
open CNID2, "> $cnid2_dump" or die "Unable to open $cnid2_dump for writing: $!";
output_header(*CNID2, "cnid2.db", scalar(@output_list) + 1);
foreach my $entry (@output_list) {
    print CNID2 " ", $entry->{cnid}, "\n";
    print CNID2 " ", $entry->{cnid}, 
                     "00000000", $entry->{dev}, "00000000", $entry->{ino},
                     $entry->{type},
                     $entry->{did}, $entry->{hexname}, "\n";
}
print CNID2 " ", $ri_cnid, "\n";
print CNID2 " ", $ri_cnid, $ri_dev, $ri_ino, $current_cnid, $ri_did, $ri_hexname, "\n";
print CNID2 "DATA=END\n";

output_header(*CNID2, "devino.db", scalar(@output_list) + 1);
foreach my $entry (@output_list) {
    print CNID2 " ", "00000000", $entry->{dev}, "00000000", $entry->{ino}, "\n";
    print CNID2 " ", $entry->{cnid}, "\n";
}
print CNID2 " ", $ri_dev, $ri_ino, "\n";
print CNID2 " ", $ri_cnid, "\n";
print CNID2 "DATA=END\n";

output_header(*CNID2, "didname.db", scalar(@output_list) + 1);
foreach my $entry (@output_list) {
    print CNID2 " ", $entry->{did}, $entry->{hexname}, "\n";
    print CNID2 " ", $entry->{cnid}, "\n";
}
print CNID2 " ", $ri_did, $ri_hexname, "\n";
print CNID2 " ", $ri_cnid, "\n";
print CNID2 "DATA=END\n";

close CNID2;

exit 0;