Transfer/Sync/Core.pm

## -------------------------------------------------------------------------
##
##    $Id: Core.pm,v 1.12 2004/06/19 01:45:08 joko Exp $
##
##    Copyright (c) 2002  Andreas Motl <andreas.motl@ilo.de>
##
##    See COPYRIGHT section in pod text below for usage and distribution rights.
##
## -------------------------------------------------------------------------
##    $Log: Core.pm,v $
##    Revision 1.12  2004/06/19 01:45:08  joko
##    introduced "local checksum"-mechanism
##    moved _dumpCompact to ::Compare::Checksum
##
##    Revision 1.11  2004/05/11 20:03:48  jonen
##    bugfix[joko] related to Attribute Map
##
##    Revision 1.10  2003/06/25 23:03:57  joko
##    no debugging
##
##    Revision 1.9  2003/05/13 08:17:52  joko
##    buildAttributeMap now propagates error
##
##    Revision 1.8  2003/03/27 15:31:15  joko
##    fixes to modules regarding new namespace(s) below Data::Mungle::*
##
##    Revision 1.7  2003/02/21 08:01:11  joko
##    debugging, logging
##    renamed module
##
##    Revision 1.6  2003/02/14 14:03:49  joko
##    + logging, debugging
##    - refactored code to sister module
##
##    Revision 1.5  2003/02/11 05:30:47  joko
##    + minor fixes and some debugging mud
##
##    Revision 1.4  2003/02/09 05:01:10  joko
##    + major structure changes
##    - refactored code to sister modules
##
##    Revision 1.3  2003/01/20 17:01:14  joko
##    + cosmetics and debugging
##    + probably refactored code to new plugin-modules 'Metadata.pm' and/or 'StorageInterface.pm' (guess it was the last one...)
##
##    Revision 1.2  2003/01/19 02:05:42  joko
##    - removed pod-documentation: now in Data/Transfer/Sync.pod
##
##    Revision 1.1  2003/01/19 01:23:04  joko
##    + new from Data/Transfer/Sync.pm
##
##    Revision 1.11  2002/12/23 07:10:59  joko
##    + using MD5 for checksum generation again - the 32-bit integer hash from DBI seems to be too lazy
##
##    Revision 1.10  2002/12/19 01:07:16  joko
##    + fixed output done via $logger
##
##    Revision 1.9  2002/12/16 07:02:34  jonen
##    + added comment
##
##    Revision 1.8  2002/12/15 02:03:09  joko
##    + fixed logging-messages
##    + additional metadata-checks
##
##    Revision 1.7  2002/12/13 21:49:34  joko
##    + sub configure
##    + sub checkOptions
##
##    Revision 1.6  2002/12/06 04:49:10  jonen
##    + disabled output-puffer here
##
##    Revision 1.5  2002/12/05 08:06:05  joko
##    + bugfix with determining empty fields (Null) with DBD::CSV
##    + debugging
##    + updated comments
##
##    Revision 1.4  2002/12/03 15:54:07  joko
##    + {import}-flag is now {prepare}-flag
##
##    Revision 1.3  2002/12/01 22:26:59  joko
##    + minor cosmetics for logging
##
##    Revision 1.2  2002/12/01 04:43:25  joko
##    + mapping detail entries may now be either an ARRAY or a HASH
##    + erase flag is used now (for export-operations)
##    + expressions to refer to values inside deep nested structures
##    - removed old mappingV2-code
##    + cosmetics
##    + sub _erase_all
##
##    Revision 1.1  2002/11/29 04:45:50  joko
##    + initial check in
##
##    Revision 1.1  2002/10/10 03:44:21  cvsjoko
##    + new
## -------------------------------------------------------------------------


package Data::Transfer::Sync::Core;

use strict;
use warnings;

use mixin::with qw( Data::Transfer::Sync );


# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -   main

use Data::Dumper;

#use misc::HashExt;
use Hash::Serializer;
use Data::Mungle::Compare::Struct qw( getDifference isEmpty );
use Data::Mungle::Transform::Deep qw( deep_copy expand );
use Data::Storage::Container;
use DesignPattern::Object;
use shortcuts::database qw( quotesql );

# get logger instance
my $logger = Log::Dispatch::Config->instance;

$| = 1;

=pod
sub new {
  my $invocant = shift;
  my $class = ref($invocant) || $invocant;
  my $self = {};
  $logger->debug( __PACKAGE__ . "->new(@_)" );
  bless $self, $class;
  $self->configure(@_);
  return $self;
}
=cut


sub _init {
  my $self = shift;

  # build new container if necessary
  $self->{container} = Data::Storage::Container->new() if !$self->{container};
  
  # add storages to container (optional)
  foreach (keys %{$self->{storages}}) {
    $self->{container}->addStorage($_, $self->{storages}->{$_});
  }
  
  # trace
    #print Dumper($self);
    #exit;

  return 1;
  
}

sub _initV1 {
  my $self = shift;
  $logger->debug( __PACKAGE__ . "->_initV1" );
  die("this should not be reached!");
  # tag storages with id-authority and checksum-provider information
  # TODO: better store tag inside metadata to hold bits together!
  map { $self->{container}->{storage}->{$_}->{isIdentAuthority} = 1 } @{$self->{id_authorities}};
  map { $self->{container}->{storage}->{$_}->{isChecksumAuthority} = 1; } @{$self->{checksum_authorities}};
  map { $self->{container}->{storage}->{$_}->{isWriteProtected} = 1; } @{$self->{write_protected}};
}


# TODO: abstract the hardwired use of "source" and "target" in here somehow - hmmmm.......  /(="§/%???
sub _run {

  my $self = shift;
  
  #print "verbose: ", $self->{verbose}, "\n";
  $self->{verbose} = 1;
  
  $logger->debug( __PACKAGE__ . "->_run" );

  # for statistics
  my $tc = Hash::Serializer->new( {} );
  my $results;
  
  # set of objects is already in $self->{args}
  # TODO: make independent of the terminology "object..."
  $results = $self->{args}->{objectSet} if $self->{args}->{objectSet};

  # apply filter
  if (my $filter = $self->{meta}->{source}->{filter}) {
    #print Dumper($filter);
    #exit;
    $results ||= $self->_getNodeList('source', $filter);
  }
  
  # get reference to node list from convenient method provided by CORE-HANDLE
  $results ||= $self->_getNodeList('source');

  # checkpoint: do we actually have a list to iterate through?
  if (!$results || !@{$results}) {
    $logger->notice( __PACKAGE__ . "->_run: No nodes to synchronize." );
    return;
  }

  #print Dumper(@$results);
  #exit;

  # check if we actually *have* a synchronization method
  if (!$self->{options}->{metadata}->{syncMethod}) {
    $logger->critical( __PACKAGE__ . "->_run: No synchronization method (checksum|timestamp) specified" );
    return;
  }
    
  
  # dereference
  my @results = @{$results};
  #print Dumper(@results);

  # iterate through set
  foreach my $source_node_real (@results) {

    print ":" if $self->{verbose};

    #print Dumper($source_node_real);

    $tc->{total}++;

    #print "=" x 80, "\n";

    # clone object (in case we have to modify it here)
    # TODO:
    #   - is a "deep_copy" needed here if occouring modifications take place?
    #   - puuhhhh, i guess a deep_copy would destroy tangram mechanisms?
    #   - after all, just take care for now that this object doesn't get updated!
    #   - so, just use its reference for now - if some cloning is needed in future, do this here!
    my $source_node = $source_node_real;
    #my $source_node = expand($source_node_real);

    # modify entry - handle new style callbacks (the readers)

    # trace
    #print Dumper($source_node);
    #exit;

    my $descent = 'source';

    # handle callbacks right now while scanning them (asymmetric to the writers)
    my $map_callbacks = {};
    if (my $callbacks = $self->{meta}->{$descent}->{Callback}) {

      # trace
        #print Dumper($callbacks);
        #exit;

      my $error = 0;

      foreach my $node (keys %{$callbacks->{read}}) {
        
        #print "cb_node: $node", "\n";
        
        my $object = $source_node;
        my $value; # = $source_node->{$node};

      # trace
        #print Dumper($self->{options});

        # ------------  half-redundant: make $self->callCallback($object, $value, $opts)
        #my $perl_callback = $self->{meta}->{$descent}->{node} . '::' . $node . '_read';
        my $perl_callback = $self->{meta}->{$descent}->{nodeType} . '::' . $node . '_read';
        my $evalstring = 'return ' . $perl_callback . '( { object => $object, property => $node, value => $value, storage => $self->{meta}->{$descent}->{storage} } );';
        #print $evalstring, "\n"; exit;
        my $cb_result = eval($evalstring);
        if ($@) {
          $error = 1;
          $logger->error( __PACKAGE__ . "->_run: $@" );
          next;
        }
        # ------------  half-redundant: make $self->callCallback($object, $value, $opts)
        
        $source_node->{$node} = $cb_result;

      }

    }

    # trace
      #print Dumper($source_node);

    # exclude defined fields  (simply delete from object)
    map { delete $source_node->{$_} } @{$self->{meta}->{source}->{subnodes_exclude}};
    
    # here we accumulate information about the status of the current node (payload/row/object/item/entry)
    $self->{node} = {};
    $self->{node}->{source}->{payload} = $source_node;

    # trace
      #print Dumper($self->{node});
      #exit;

    # determine ident of entry
    my $identOK = $self->_resolveNodeIdent('source');
    #if (!$identOK && lc $self->{args}->{direction} ne 'import') {
    if (!$identOK) {
      #print Dumper($self->{meta}->{source});
      $logger->critical( __PACKAGE__ . "->_run: No ident found in source node ( nodeName='$self->{meta}->{source}->{nodeName}', nodeType='$self->{meta}->{source}->{nodeType}') try to \"prepare\" this node first?" );
      return;
    }

    #print "l" if $self->{verbose};
    my $statOK = $self->_statloadNode('target', $self->{node}->{source}->{ident});

    # mark node as new either if there's no ident or if stat/load failed
    if (!$statOK) {
      $self->{node}->{status}->{new} = 1;
      print "n" if $self->{verbose};
    }

    # determine status of entry by synchronization method
    if ( lc $self->{options}->{metadata}->{syncMethod} eq 'checksum' ) {
    #if ( $statOK && (lc $self->{args}->{method} eq 'checksum') ) {
    #if ( !$self->{node}->{status}->{new} && (lc $self->{args}->{method} eq 'checksum') ) {
      
    # new 2004-06-17: calculate local checksum of source node
    # new 2004-06-19: ... only if requested
    if ($self->{options}->{source}->{storage}->{handle}->{locator}->{hasLocalChecksum}) {
      $self->handleLocalChecksum('source');
    }
    
    # calculate checksum of source node
      #$self->_calcChecksum('source');
      if (!$self->readChecksum('source')) {
        $logger->warning( __PACKAGE__ . "->_run: Could not find \"source\" entry with ident=\"$self->{node}->{source}->{ident}\"" );
        $tc->{skip}++;
        print "s" if $self->{verbose};
        next;
      }
      
      # get checksum from synchronization target
      $self->readChecksum('target');
      #if (!$self->readChecksum('target')) {
      #  $logger->critical( __PACKAGE__ . "->readChecksum: Could not find \"target\" entry with ident=\"$self->{node}->{source}->{ident}\"" );
      #  next;
      #}
  
      # pre flight check: do we actually have a checksum provided?
      #if (!$self->{node}->{source}->{checksum}) {
      #  print "Source checksum for entry with ident \"$self->{node}->{source}->{ident}\" could not be calculated, maybe it's missing?.", "\n";
      #  return;
      #}
      
      # determine if entry is "new" or "dirty"
      # after all, this seems to be the point where the hammer falls.....
      print "c" if $self->{verbose};

      # trace
        #print Dumper($self->{node});
        #exit;
        #print "LOCAL: ", $self->{node}->{source}->{checksum_local_storage}, " <-> ", $self->{node}->{source}->{checksum_local_calculated}, "\n";
        #print "REMOTE: ", $self->{node}->{source}->{checksum}, " <-> ", $self->{node}->{target}->{checksum}, "\n";

      # calculate new/dirty status
      $self->{node}->{status}->{new} = !$self->{node}->{target}->{checksum};
      if (!$self->{node}->{status}->{new}) {
        $self->{node}->{status}->{dirty} = 
          $self->{node}->{status}->{new} || 
          (!$self->{node}->{source}->{checksum} || !$self->{node}->{target}->{checksum}) || 
          ($self->{node}->{source}->{checksum} ne $self->{node}->{target}->{checksum}) || 
          $self->{args}->{force};
      }

      # new 2004-06-17: also check if local checksum is inconsistent
      # new 2004-06-19: ... only if requested
      if ($self->{options}->{source}->{storage}->{handle}->{locator}->{hasLocalChecksum} and
        ($self->{node}->{source}->{checksum_local_storage} ne $self->{node}->{source}->{checksum_local_calculated}) ) {
        $self->{node}->{status}->{dirty_local} = 1;
        $self->{node}->{status}->{dirty} = 1;
      }

    } else {
      $logger->warning( __PACKAGE__ . "->_run: Synchronization method '$self->{options}->{metadata}->{syncMethod}' is not implemented" );
      $tc->{skip}++;
      print "s" if $self->{verbose};
      next;
    }

    # new 2004-06-17: also update local checksum
    if ($self->{node}->{status}->{dirty_local}) {
      $tc->{locally_modified}++;
      print "[lm]" if $self->{verbose};
      $self->_doModify_LocalChecksum('source');
    }

    # first reaction on entry-status: continue with next entry if the current is already "in sync"
    if (!$self->{node}->{status}->{new} && !$self->{node}->{status}->{dirty}) {
      $tc->{in_sync}++;
      next;
    }

    #print Dumper($self->{node}->{source});

    # build map to actually transfer the data from source to target
    if (!$self->buildAttributeMap()) {
      #$logger->warning( __PACKAGE__ . "->_run: Attribute Map could not be created. Will not insert or modify node.");
      push( @{$tc->{error_per_row}}, "Attribute Map could not be created. Will not insert or modify node $self->{node}->{source}->{ident}.");
      #push( @{$tc->{error_per_row}}, "Attribute Map could not be created. Will not insert or modify node " . Dumper($self->{node}->{source}) . ".");
      $tc->{error}++;
      print "e" if $self->{verbose};
      next;
    }

    # trace
      #print Dumper($self->{node}); exit;
      #print "attempt", "\n";

    # additional (new) checks for feature "write-protection"
    if ($self->{meta}->{target}->{storage}->{isWriteProtected}) {
      $tc->{attempt_transfer}++;
      print "\n" if $self->{verbose};
      $logger->notice( __PACKAGE__ . "->_run: Target is write-protected. Will not insert or modify node. " .
          "(Ident: $self->{node}->{source}->{ident} " . "Dump:\n" . Dumper($self->{node}->{source}->{payload}) . ")" );
      print "\n" if $self->{verbose};
      $tc->{skip}++;
      next;
    }

    # trace
      #print Dumper($self);
      #exit;

    # transfer contents of map to target
    if ($self->{node}->{status}->{new}) {
      $tc->{attempt_new}++;
      $self->_doTransferToTarget('insert');
      # asymmetry: refetch node from target to re-calculate new ident and checksum (TODO: is IdentAuthority of relevance here?)
      #print Dumper($self->{node});
      $self->_statloadNode('target', $self->{node}->{target}->{ident}, 1);
      $self->readChecksum('target');

    } elsif ($self->{node}->{status}->{dirty}) {
      $tc->{attempt_modify}++;
      # asymmetry: get ident before updating (TODO: is IdentAuthority of relevance here?)
      $self->{node}->{target}->{ident} = $self->{node}->{map}->{$self->{meta}->{target}->{IdentProvider}->{arg}};
      $self->_doTransferToTarget('update');
      $self->readChecksum('target');
    }

    if ($self->{node}->{status}->{ok}) {
      $tc->{ok}++;
      print "t" if $self->{verbose};
    }
    
    if ($self->{node}->{status}->{error}) {
      $tc->{error}++;
      push( @{$tc->{error_per_row}}, $self->{node}->{status}->{error} );
      print "e" if $self->{verbose};
    }
    
    # trace
      #print Dumper($self);
      #exit;
    
    # change ident in source (take from target), if transfer was ok and target is an IdentAuthority
    # this is (for now) called a "retransmit" indicated by a "r"-character when verbosing
    #if ($self->{node}->{status}->{ok} && $self->{options}->{target}->{storage}->{idAuthority}) {
    if ($self->{node}->{status}->{ok} && $self->{meta}->{target}->{isIdentAuthority}) {
      print "r" if $self->{verbose};
      #print Dumper($self->{meta});
      #print Dumper($self->{node});
      #exit;
      $self->_doModifySource_IdentChecksum($self->{node}->{target}->{ident});
    }
    
    #print "UNLOAD", "\n";
    #$self->{meta}->{source}->{storage}->unload( $self->{node}->{source}->{payload} );

  }

  print "\n" if $self->{verbose};
  
  # build user-message from some stats
    my $msg = "statistics: $tc";
    
    if ($tc->{error_per_row}) {
      $msg .= "\n";
      $msg .= "errors from \"error_per_row\":" . "\n";
      $msg .= Dumper($tc->{error_per_row});
    }
    
    # todo!!!
    #sysevent( { usermsg => $msg, level => $level }, $taskEvent );
    #$logger->info( __PACKAGE__ . "->_run: $msg" );
    $logger->info($msg . "\n");

  return $tc;

}


sub _doTransferToTarget {
  my $self = shift;
  my $action = shift;
    
  # trace
    #print Dumper($self->{meta});
    #print Dumper($self->{node});
    #exit;
    
  $self->_modifyNode('target', $action, $self->{node}->{map});
}


sub _doModifySource_IdentChecksum {
  my $self = shift;
  my $ident_new = shift;
  # this changes an old node to a new one including ident and checksum
  # TODO:
  #   - eventually introduce an external resource to store this data to
  #   - we won't have to "re"-modify the source node here
  my $map = {
    $self->{meta}->{source}->{IdentProvider}->{arg} => $ident_new,
    cs => $self->{node}->{target}->{checksum},
  };

#print Dumper($map);
#print Dumper($self->{node});
#exit;

  $self->_modifyNode('source', 'update', $map);
}

sub _doModify_LocalChecksum {
  my $self = shift;
  my $descent = shift;
  my $map = {
    cs_local => $self->{node}->{$descent}->{checksum_local_calculated},
  };
  $self->_modifyNode($descent, 'update', $map);
}

sub _prepareNode_MetaProperties {
  my $self = shift;
  my $descent = shift;

  $logger->info( __PACKAGE__ . "->_prepareNode_MetaProperties( descent $descent )" );
  
  # TODO: this should (better) be: "my $firstnode = $self->_getFirstNode($descent);"
  my $list = $self->_getNodeList($descent);
  
  # get first node
  my $firstnode = $list->[0];
  
  # check if node contains meta properties/nodes
  # TODO: "cs" is hardcoded here!
  my @required = ( $self->{meta}->{$descent}->{IdentProvider}->{arg}, 'cs' );
  my @found = keys %$firstnode;
  #my @diff = getDifference(\@found, \@required);
  my $diff = getDifference(\@required, \@found);
  #print Dumper(@found);
  #print Dumper(@required);
  #print Dumper(@diff);
  #if (!$#diff || $#diff == -1) {
  if (isEmpty($diff)) {
    $logger->warning( __PACKAGE__ . "->_prepareNode_MetaProperties: node is lacking meta properties - will try to alter..." );
    foreach (@required) {
      my $sql = "ALTER TABLE $self->{meta}->{$descent}->{node} ADD COLUMN $_";
      #print "sql: $sql", "\n";
      my $res = $self->{meta}->{$descent}->{storage}->sendCommand($sql);
      #print Dumper($res->getStatus());
    }
  }

}

# TODO: load column-metadata from reversed mapping-metadata
sub _prepareNode_DummyIdent {
  my $self = shift;
  my $descent = shift;

  #print Dumper($self->{options});
  $logger->info( __PACKAGE__ . "->_prepareNode_DummyIdent( descent $descent )" );

  my $list = $self->_getNodeList($descent);
  #print Dumper($list);
  my $i = 0;
  my $ident_base = 5678983;
  my $ident_appendix = '0001';
  foreach my $node (@$list) {
    my $ident_dummy = $i + $ident_base;
    $ident_dummy .= $ident_appendix;
    my $map = {
      $self->{meta}->{$descent}->{IdentProvider}->{arg} => $ident_dummy,
      cs => undef,
      cs_local => undef,
    };
    
    # diff lists and ...
    my $diff = getDifference([keys %$node], [keys %$map]);
    next if $#{$diff} == -1;
    
    # ... build criteria including all columns
    my @crits;
    foreach my $property (@$diff) {
      next if !$property;
      my $value = $node->{$property};
      next if !$value;
      push @crits, "$property='" . quotesql($value) . "'";
    }
    my $crit = join ' AND ', @crits;
    print "p" if $self->{verbose};

#print Dumper($map);
#print Dumper($crit);

    $self->_modifyNode($descent, 'update', $map, $crit);
    $i++;
  }

  print "\n" if $self->{verbose};
  
  if (!$i) {
    $logger->warning( __PACKAGE__ . "->_prepareNode_DummyIdent: no nodes touched" );
  }
  
}

# TODO: handle this in an abstract way (wipe out use of 'source' and/or 'target' inside core)
sub _otherSide {
  my $self = shift;
  my $descent = shift;
  return 'source' if $descent eq 'target';
  return 'target' if $descent eq 'source';
  return '';
}


1;
__END__
MailToCvsAdmin">MailToCvsAdmin	ViewVC Help
Powered by ViewVC 1.1.26