#!/usr/bin/perl -w

use strict;
use DBI;
use Data::Dumper;
use Data::TreeDumper;

# Define variables
my $db;
my $user;
my $pass;
my $host;
my $dbh;
my $stime;
my $etime;
my $endtime;
my %hash;

## DB data
$db		= "ofrs";
$user	= "ofrs";
$pass	= "ofrs";
$host	= "localhost";

# Connect to DB
$dbh = DBI->connect("DBI:mysql:$db:$host", $user, $pass);

# Set time before import
$stime = time;

# Sync a replication folder on the disk with the database
checkFS("/data/test1"); 

# Set time after import
$etime = time;
$endtime = ($etime - $stime);

# Print needed time
print "Needed $endtime seconds ...\n";

# Creates a hash tree for a replication folder using the pathes in the database
# Syntax: makeTree($repFolder)
# Return Value: hash-tree (hash) 
sub makeTree {
  # Set & define the variables
  my $path = $_[0];
  my $tree={};
  
  # Set query
  my $query = "SELECT path FROM dir WHERE path = '$path';";

  # Run query
  my $sth = $dbh->prepare($query);
  $sth->execute();

  # Use the query results ...
  while (my @result = $sth->fetchrow_array) {
    my $path = $result[0];
    
    # ... and add each path to the hash-tree
    addToTree($tree,$path);  
       
  }

  # return the hash-tree
  return $tree;  
    
}

# add a Path to the hash-tree
# Syntax: addToTree($tree,$path)
# Return value: none
sub addToTree {
  # Set & define the variables
  my $tree = $_[0];
  my $path = $_[1];
  
  # Split the path up to its parts  
  my @parts = split('/',$path);

  # Delete the first (empty) part  
  shift @parts;

  # Delete the trailing slash at the end of the path if exists  
  my $file=1;
  unless($parts[-1])
  {
    $file=0;
    pop(@parts);
  }
  
  # build the hash-tree for this path
  my $ref=\$tree;
  $ref=\$$ref->{$_} for(@parts);

}

# build the pathes from the hash-tree
# Syntax: buildPathes($tree)
# Return value: path-list (array-reference)
sub buildPathes
{
  # Set & define variables
  my $tree=shift;
  my $path=shift // '';
  my @list;
  
  # walk through all levels and sort them
  for my $name (sort(keys(%$tree)))
  {
    my $elm=$tree->{$name};

    if($elm && ref($elm) eq 'HASH')
    {
      
      # run the function recursive
      my $lst=buildPathes($elm,"$path/$name");
      push(@list,@$lst);
    }
    else
    { push(@list,"$path/$name"); }
  }

  # return the list of pathes from the hash-tree
  return \@list;
}

# sync a replication folder on the disk with the database
# Syntax: checkFS($repPath)
# Return value: none
sub checkFS {
  # Set & define variables
  my $path = $_[0];

  # create the hash-tree for this path / replication folder 
  my $tree = makeTree($path);

  # check if the folder on the disk exists in the database
  checkFolder($path, $tree);

  # build the pathes from the hashes left in the hash-tree  
  my $onlyDB = buildPathes($tree);

  # print the pathes out  
  print "OnlyDB-Pathes:\n";
  print Dumper($onlyDB);
  
#  foreach my $path (@$onlyDB) {
#    DELETE THE MISSING FOLDERS FROM THE DATABASE
#  }
  
}

# check a folder and its subfolder if they exists in the hash-tree
# Syntax: checkFolder($path, $tree)
# Return value: none
sub checkFolder {
  # Set variables
	my @subfolders;
  my $path;
  my @parts;
  my $name;
  my $hash;
  my $tmphash;
  my $found;
  my $childNotFound;

  # Load arguments into variables
	if (-d $_[0]) {
    $path = $_[0]; # Load path into variable
  } else {
    # Exit script if no path is specified
    print "Error: No valid path specified\n";
    exit(1);    
  }
  if (defined($_[1])) {
    $hash = $_[1]; # Load hash into variable
  } else {
    # Exit script if no hash is specified
    print "Error: No hash specified\n";
    exit(1);    
  }

  # Set variables
  $childNotFound = 0;

  # split the path to its parts
	@parts=split(/\//, $path);
	shift @parts;
	$name = $parts[-1];
	$found = 0;
	
	$tmphash = \%hash;
	for (my $i = 0; $i < @parts; $i++) { # walk through each part of the path
	 my $part = $parts[$i];
		if(defined($tmphash->{$part}))  { # Does the part exists in the hash-tree?
			if($i == @parts-1) { # Is this the last part of the path?
				$found=1; # Yes, the folder is in the database!
			} else { # Not the last part of the path
				$tmphash=$tmphash->{$part}; # Go one level further
			}
		} else {		  
			last; # The path is not in the database
		}
	}  

  if ($found == 0) {
    # Insert current directory into DB
    # Set query
    my $query = "INSERT INTO dir (path) VALUES (\"$path\");";
  
    # Run query
    $dbh->do($query);    
  }

  # Open the current directory  
  opendir (DIR, $path) or die "Unable to open $path: $!";

  # Exclude "." and ".."
  my @files = grep { !/^\.{1,2}$/ } readdir (DIR);

  # Close directory
  closedir (DIR);
  
  # Set the full path for all files
  @files = map { $path . '/' . $_ } @files;


  # Write each subfolder into the array @subfolders
  foreach my $file (@files) {
    if (-d $file) {
      push (@subfolders, $file)
    }  
  }
  
  # If there are any subfolders then ...
  if (@subfolders > 0) {
    foreach my $subfolder (@subfolders) {
      # ... put the subfolder into the DB using the subfolder's path and the id of the current directory
      $childNotFound = checkFolder($subfolder, \%hash);
    }    
  } 
  
  # if this directory was found in the path and if it has no subfolders, then delete it from the hash-tree
  if ($found == 1 && !keys(%{$tmphash->{$name}})) {
    delete $tmphash->{$name};
    return 0;
  } else {
    # return 1 if this folder is either not in the hash-tree or if this directory has subfolders
    return 1;
  }
    
}