1 |
joko |
1.1 |
## $Id: StorageInterface.pm,v 1.1 2003/01/20 16:58:46 joko Exp $ |
2 |
|
|
## |
3 |
|
|
## Copyright (c) 2002 Andreas Motl <andreas.motl@ilo.de> |
4 |
|
|
## |
5 |
|
|
## See COPYRIGHT section in pod text below for usage and distribution rights. |
6 |
|
|
## |
7 |
|
|
## ---------------------------------------------------------------------------------------- |
8 |
|
|
## $Log: StorageInterface.pm,v $ |
9 |
|
|
## ---------------------------------------------------------------------------------------- |
10 |
|
|
|
11 |
|
|
|
12 |
|
|
package Data::Transfer::Sync::Compare::Checksum; |
13 |
|
|
|
14 |
|
|
use strict; |
15 |
|
|
use warnings; |
16 |
|
|
|
17 |
|
|
use mixin::with qw( Data::Transfer::Sync ); |
18 |
|
|
|
19 |
|
|
|
20 |
|
|
# - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - main |
21 |
|
|
|
22 |
|
|
use Data::Dumper; |
23 |
|
|
use Digest::MD5 qw(md5 md5_hex md5_base64); |
24 |
|
|
|
25 |
|
|
|
26 |
|
|
# get logger instance |
27 |
|
|
my $logger = Log::Dispatch::Config->instance; |
28 |
|
|
|
29 |
|
|
|
30 |
|
|
sub _calcChecksum { |
31 |
|
|
|
32 |
|
|
my $self = shift; |
33 |
|
|
my $descent = shift; |
34 |
|
|
my $specifier = shift; |
35 |
|
|
|
36 |
|
|
# calculate checksum for current object |
37 |
|
|
my $ident = $self->{node}->{$descent}->{ident}; |
38 |
|
|
|
39 |
|
|
# build dump of this node |
40 |
|
|
my $payload = $self->{node}->{$descent}->{payload}; |
41 |
|
|
#my $dump = $ident . "\n" . $item->quickdump(); |
42 |
|
|
#my $dump = $ident . "\n" . Dumper($item); |
43 |
|
|
my $dump = $ident . "\n" . $self->_dumpCompact($payload); |
44 |
|
|
|
45 |
|
|
# TODO: $logger->dump( ... ); |
46 |
|
|
#$logger->debug( __PACKAGE__ . ": " . $dump ); |
47 |
|
|
#$logger->dump( __PACKAGE__ . ": " . $dump ); |
48 |
|
|
|
49 |
|
|
# calculate checksum from dump |
50 |
|
|
# note: the 32-bit integer hash from DBI seems |
51 |
|
|
# to generate duplicates with small payloads already in ranges of hundreds of items/rows!!! |
52 |
|
|
# try to avoid to use it or try to use it only for payloads greater than, hmmm, let's say 30 chars? |
53 |
|
|
# (we had about 15 chars average per item (row)) |
54 |
|
|
|
55 |
|
|
# md5-based fingerprint, base64 encoded (from Digest::MD5) |
56 |
|
|
$self->{node}->{$descent}->{checksum} = md5_base64($dump) . '=='; |
57 |
|
|
# 32-bit integer "hash" value (maybe faster?) (from DBI) |
58 |
|
|
#$self->{node}->{$descent}->{checksum} = DBI::hash($dump, 1); |
59 |
|
|
|
60 |
|
|
# signal good |
61 |
|
|
return 1; |
62 |
|
|
|
63 |
|
|
} |
64 |
|
|
|
65 |
|
|
|
66 |
|
|
sub _readChecksum { |
67 |
|
|
my $self = shift; |
68 |
|
|
|
69 |
|
|
my $descent = shift; |
70 |
|
|
|
71 |
|
|
# signal checksum bad |
72 |
|
|
if (!$self->{node}->{$descent}) { |
73 |
|
|
return; |
74 |
|
|
} |
75 |
|
|
|
76 |
|
|
# trace |
77 |
|
|
#print "desc: $descent", "\n"; |
78 |
|
|
#print Dumper($self); |
79 |
|
|
#exit; |
80 |
|
|
|
81 |
|
|
# get checksum for current entry |
82 |
|
|
# TODO: don't have the checksum column/property hardcoded as "cs" here, make this configurable somehow |
83 |
|
|
if ($self->{options}->{$descent}->{storage}->{isChecksumAuthority}) { |
84 |
|
|
$self->_calcChecksum($descent); |
85 |
|
|
} else { |
86 |
|
|
$self->{node}->{$descent}->{checksum} = $self->{node}->{$descent}->{payload}->{cs}; |
87 |
|
|
} |
88 |
|
|
|
89 |
|
|
# signal checksum good |
90 |
|
|
return 1; |
91 |
|
|
|
92 |
|
|
} |
93 |
|
|
|
94 |
|
|
|
95 |
|
|
1; |