3 |
use warnings; |
use warnings; |
4 |
use strict; |
use strict; |
5 |
|
|
6 |
use WebPAC::Common; |
use WebPAC::Input; |
|
use base qw/WebPAC::Input WebPAC::Common/; |
|
7 |
|
|
8 |
=head1 NAME |
=head1 NAME |
9 |
|
|
10 |
WebPAC::Input::ISIS - support for CDS/ISIS source files |
WebPAC::Input::ISIS - support for CDS/ISIS database files |
11 |
|
|
12 |
=head1 VERSION |
=head1 VERSION |
13 |
|
|
14 |
Version 0.01 |
Version 0.04 |
15 |
|
|
16 |
=cut |
=cut |
17 |
|
|
18 |
our $VERSION = '0.01'; |
our $VERSION = '0.04'; |
19 |
|
|
20 |
|
|
|
# auto-configure |
|
|
|
|
|
my ($have_biblio_isis, $have_openisis) = (0,0); |
|
|
|
|
|
eval "use Biblio::Isis 0.13;"; |
|
|
unless ($@) { |
|
|
$have_biblio_isis = 1 |
|
|
} else { |
|
|
eval "use OpenIsis;"; |
|
|
$have_openisis = 1 unless ($@); |
|
|
} |
|
|
|
|
21 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
22 |
|
|
23 |
Open CDS/ISIS, WinISIS or IsisMarc database using Biblio::Isis or OpenIsis |
Open CDS/ISIS, WinISIS or IsisMarc database using C<Biblio::Isis> or |
24 |
module and read all records to memory. |
C<OpenIsis> module and read all records to memory. |
25 |
|
|
26 |
my $isis = new WebPAC::Input::ISIS(); |
my $isis = new WebPAC::Input::ISIS(); |
27 |
$isis->open( filename => '/path/to/ISIS/ISIS' ); |
$isis->open( path => '/path/to/ISIS/ISIS' ); |
28 |
|
|
29 |
=head1 FUNCTIONS |
=head1 FUNCTIONS |
30 |
|
|
31 |
=head2 open |
=head2 init |
32 |
|
|
33 |
This function will read whole database in memory and produce lookups. |
Autoconfigure this module to use C<Biblio::Isis> or C<OpenIsis>. |
34 |
|
|
35 |
|
=cut |
36 |
|
|
37 |
|
sub init { |
38 |
|
my $self = shift; |
39 |
|
|
40 |
$isis->open( |
eval "use Biblio::Isis;"; |
41 |
filename => '/data/ISIS/ISIS', |
unless ($@) { |
42 |
code_page => '852', |
$self->{have_biblio_isis} = 1 |
43 |
limit_mfn => 500, |
} else { |
44 |
start_mfn => 6000, |
eval "use OpenIsis;"; |
45 |
lookup => $lookup_obj, |
$self->{have_openisis} = 1 unless ($@); |
46 |
); |
} |
47 |
|
} |
48 |
|
|
49 |
By default, ISIS code page is assumed to be C<852>. |
=head2 open_db |
50 |
|
|
51 |
If optional parametar C<start_mfn> is set, this will be first MFN to read |
Returns handle to database and size in records |
|
from database (so you can skip beginning of your database if you need to). |
|
52 |
|
|
53 |
If optional parametar C<limit_mfn> is set, it will read just 500 records |
my ($db,$size) = $isis->open_db( |
54 |
from database in example above. |
path => '/path/to/LIBRI' |
55 |
|
filter => sub { |
56 |
|
my ($l,$field_nr) = @_; |
57 |
|
# do something with $l which is line of input file |
58 |
|
return $l; |
59 |
|
}, |
60 |
|
} |
61 |
|
|
62 |
Returns number of last record read into memory (size of database, really). |
Options: |
63 |
|
|
64 |
=cut |
=over 4 |
65 |
|
|
66 |
sub open { |
=item path |
|
my $self = shift; |
|
|
my $arg = {@_}; |
|
67 |
|
|
68 |
my $log = $self->_get_logger(); |
path to CDS/ISIS database |
69 |
|
|
70 |
$log->logcroak("need filename") if (! $arg->{'filename'}); |
=back |
|
my $code_page = $arg->{'code_page'} || '852'; |
|
71 |
|
|
72 |
$log->logdie("can't find database ",$arg->{'filename'}) unless (glob($arg->{'filename'}.'.*')); |
=cut |
73 |
|
|
74 |
# store data in object |
sub open_db { |
75 |
$self->{'isis_filename'} = $arg->{'filename'}; |
my $self = shift; |
|
$self->{'isis_code_page'} = $code_page; |
|
76 |
|
|
77 |
#$self->{'isis_code_page'} = $code_page; |
my $arg = {@_}; |
78 |
|
|
79 |
# create Text::Iconv object |
my $log = $self->_get_logger(); |
|
my $cp = Text::Iconv->new($code_page,$self->{'code_page'}); |
|
80 |
|
|
81 |
$log->info("reading ISIS database '",$arg->{'filename'},"'"); |
$log->info("opening ISIS database '$arg->{path}'"); |
|
$log->debug("isis code page: $code_page"); |
|
82 |
|
|
83 |
my ($isis_db,$maxmfn); |
my ($isis_db,$db_size); |
84 |
|
|
85 |
if ($have_openisis) { |
if ($self->{have_openisis}) { |
86 |
$log->debug("using OpenIsis perl bindings"); |
$log->debug("using OpenIsis perl bindings"); |
87 |
$isis_db = OpenIsis::open($arg->{'filename'}); |
$isis_db = OpenIsis::open($arg->{path}); |
88 |
$maxmfn = OpenIsis::maxRowid( $isis_db ) || 1; |
$db_size = OpenIsis::maxRowid( $isis_db ) || 1; |
89 |
} elsif ($have_biblio_isis) { |
} elsif ($self->{have_biblio_isis}) { |
90 |
$log->debug("using Biblio::Isis"); |
$log->debug("using Biblio::Isis"); |
91 |
use Biblio::Isis; |
use Biblio::Isis; |
92 |
$isis_db = new Biblio::Isis( |
$isis_db = new Biblio::Isis( |
93 |
isisdb => $arg->{'filename'}, |
isisdb => $arg->{path}, |
94 |
include_deleted => 1, |
include_deleted => 1, |
95 |
hash_filter => sub { |
hash_filter => $arg->{filter} ? sub { return $arg->{filter}->(@_); } : undef, |
96 |
my $l = shift || return; |
) or $log->logdie("can't find database $arg->{path}"); |
|
$l = $cp->convert($l); |
|
|
return $l; |
|
|
}, |
|
|
); |
|
|
$maxmfn = $isis_db->count; |
|
|
|
|
|
unless ($maxmfn) { |
|
|
$log->logwarn("no records in database ", $arg->{'filename'}, ", skipping..."); |
|
|
return; |
|
|
} |
|
|
|
|
|
} else { |
|
|
$log->logdie("Can't find supported ISIS library for perl. I suggent that you install Bilbio::Isis from CPAN."); |
|
|
} |
|
97 |
|
|
98 |
|
$db_size = $isis_db->count; |
99 |
|
|
|
my $startmfn = 1; |
|
|
|
|
|
if (my $s = $self->{'start_mfn'}) { |
|
|
$log->info("skipping to MFN $s"); |
|
|
$startmfn = $s; |
|
100 |
} else { |
} else { |
101 |
$self->{'start_mfn'} = $startmfn; |
$log->logdie("Can't find supported ISIS library for perl. I suggent that you install Bilbio::Isis from CPAN."); |
|
} |
|
|
|
|
|
$maxmfn = $startmfn + $self->{limit_mfn} if ($self->{limit_mfn}); |
|
|
|
|
|
$log->info("processing ",($maxmfn-$startmfn)." records using ",( $have_openisis ? 'OpenIsis' : 'Biblio::Isis')); |
|
|
|
|
|
|
|
|
# read database |
|
|
for (my $mfn = $startmfn; $mfn <= $maxmfn; $mfn++) { |
|
|
|
|
|
$log->debug("mfn: $mfn\n"); |
|
|
|
|
|
my $rec; |
|
|
|
|
|
if ($have_openisis) { |
|
|
|
|
|
# read record using OpenIsis |
|
|
my $row = OpenIsis::read( $isis_db, $mfn ); |
|
|
foreach my $k (keys %{$row}) { |
|
|
if ($k ne "mfn") { |
|
|
foreach my $l (@{$row->{$k}}) { |
|
|
$l = $cp->convert($l); |
|
|
# has subfields? |
|
|
my $val; |
|
|
if ($l =~ m/\^/) { |
|
|
foreach my $t (split(/\^/,$l)) { |
|
|
next if (! $t); |
|
|
$val->{substr($t,0,1)} = substr($t,1); |
|
|
} |
|
|
} else { |
|
|
$val = $l; |
|
|
} |
|
|
|
|
|
push @{$rec->{$k}}, $val; |
|
|
} |
|
|
} else { |
|
|
push @{$rec->{'000'}}, $mfn; |
|
|
} |
|
|
} |
|
|
|
|
|
} elsif ($have_biblio_isis) { |
|
|
$rec = $isis_db->to_hash($mfn); |
|
|
} else { |
|
|
$log->logdie("hum? implementation missing?"); |
|
|
} |
|
|
|
|
|
$log->confess("record $mfn empty?") unless ($rec); |
|
|
|
|
|
# store |
|
|
if ($self->{'low_mem'}) { |
|
|
$self->{'db'}->put($mfn, $rec); |
|
|
} else { |
|
|
$self->{'data'}->{$mfn} = $rec; |
|
|
} |
|
|
|
|
|
# create lookup |
|
|
$self->{'lookup'}->add( $rec ) if ($self->{'lookup'} && can($self->{'lookup'}->add)); |
|
|
|
|
|
$self->progress_bar($mfn,$maxmfn); |
|
|
|
|
102 |
} |
} |
103 |
|
|
104 |
$self->{'current_mfn'} = -1; |
return ($isis_db, $db_size); |
|
$self->{'last_pcnt'} = 0; |
|
|
|
|
|
$log->debug("max mfn: $maxmfn"); |
|
|
|
|
|
# store max mfn and return it. |
|
|
return $self->{'max_mfn'} = $maxmfn; |
|
105 |
} |
} |
106 |
|
|
107 |
=head2 fetch_rec |
=head2 fetch_rec |
108 |
|
|
109 |
Fetch next record from database. It will also displays progress bar. |
Return record with ID C<$mfn> from database |
110 |
|
|
111 |
my $rec = $webpac->fetch_rec; |
my $rec = $self->fetch_rec( $db, $mfn ); |
112 |
|
|
113 |
=cut |
=cut |
114 |
|
|
115 |
sub fetch_rec { |
sub fetch_rec { |
116 |
my $self = shift; |
my $self = shift; |
117 |
|
|
118 |
my $log = $self->_get_logger(); |
my ($isis_db, $mfn) = @_; |
|
|
|
|
$log->logconfess("it seems that you didn't load database!") unless ($self->{'current_mfn'}); |
|
119 |
|
|
120 |
if ($self->{'current_mfn'} == -1) { |
my $rec; |
|
$self->{'current_mfn'} = $self->{'start_mfn'}; |
|
|
} else { |
|
|
$self->{'current_mfn'}++; |
|
|
} |
|
121 |
|
|
122 |
my $mfn = $self->{'current_mfn'}; |
if ($self->{have_openisis}) { |
123 |
|
|
124 |
if ($mfn > $self->{'max_mfn'}) { |
# read record using OpenIsis |
125 |
$self->{'current_mfn'} = $self->{'max_mfn'}; |
my $row = OpenIsis::read( $isis_db, $mfn ); |
|
$log->debug("at EOF"); |
|
|
return; |
|
|
} |
|
126 |
|
|
127 |
$self->progress_bar($mfn,$self->{'max_mfn'}); |
# convert record to hash |
128 |
|
foreach my $k (keys %{$row}) { |
129 |
|
if ($k ne "mfn") { |
130 |
|
foreach my $l (@{$row->{$k}}) { |
131 |
|
$l = $self->{iconv}->convert($l) if ($self->{iconv}); |
132 |
|
# has subfields? |
133 |
|
my $val; |
134 |
|
if ($l =~ m/\^/) { |
135 |
|
foreach my $t (split(/\^/,$l)) { |
136 |
|
next if (! $t); |
137 |
|
$val->{substr($t,0,1)} = substr($t,1); |
138 |
|
} |
139 |
|
} else { |
140 |
|
$val = $l; |
141 |
|
} |
142 |
|
push @{$rec->{"$k"}}, $val; |
143 |
|
} |
144 |
|
} else { |
145 |
|
push @{$rec->{'000'}}, $mfn; |
146 |
|
} |
147 |
|
} |
148 |
|
|
149 |
if ($self->{'low_mem'}) { |
} elsif ($self->{have_biblio_isis}) { |
150 |
return $self->{'db'}->get($mfn); |
$rec = $isis_db->to_hash({ |
151 |
|
mfn => $mfn, |
152 |
|
include_subfields => 1 |
153 |
|
}); |
154 |
} else { |
} else { |
155 |
return $self->{'data'}->{$mfn}; |
$self->_get_logger()->logdie("hum? implementation missing?"); |
156 |
} |
} |
|
} |
|
|
|
|
|
=head2 mfn |
|
|
|
|
|
Returns current record number (MFN). |
|
157 |
|
|
158 |
print $webpac->mfn; |
return $rec; |
|
|
|
|
=cut |
|
|
|
|
|
sub mfn { |
|
|
my $self = shift; |
|
|
return $self->{'current_mfn'}; |
|
159 |
} |
} |
160 |
|
|
161 |
=head1 AUTHOR |
=head1 AUTHOR |