3 |
use warnings; |
use warnings; |
4 |
use strict; |
use strict; |
5 |
|
|
|
use blib; |
|
|
|
|
6 |
use WebPAC::Common; |
use WebPAC::Common; |
7 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
8 |
use Text::Iconv; |
use Text::Iconv; |
9 |
|
use Data::Dumper; |
10 |
|
|
11 |
=head1 NAME |
=head1 NAME |
12 |
|
|
14 |
|
|
15 |
=head1 VERSION |
=head1 VERSION |
16 |
|
|
17 |
Version 0.03 |
Version 0.04 |
18 |
|
|
19 |
=cut |
=cut |
20 |
|
|
21 |
our $VERSION = '0.03'; |
our $VERSION = '0.04'; |
22 |
|
|
23 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
24 |
|
|
47 |
); |
); |
48 |
|
|
49 |
$db->open('/path/to/database'); |
$db->open('/path/to/database'); |
50 |
print "database size: ",$db->size,"\n"; |
print "database size: ",$db->size,"\n"; |
51 |
while (my $rec = $db->fetch) { |
while (my $rec = $db->fetch) { |
52 |
} |
# do something with $rec |
53 |
|
} |
54 |
|
|
55 |
|
|
56 |
|
|
64 |
module => 'WebPAC::Input::MARC', |
module => 'WebPAC::Input::MARC', |
65 |
code_page => 'ISO-8859-2', |
code_page => 'ISO-8859-2', |
66 |
low_mem => 1, |
low_mem => 1, |
67 |
|
recode => 'char pairs', |
68 |
|
no_progress_bar => 1, |
69 |
); |
); |
70 |
|
|
71 |
C<module> is low-level file format module. See L<WebPAC::Input::Isis> and |
C<module> is low-level file format module. See L<WebPAC::Input::Isis> and |
77 |
|
|
78 |
Default is not to use C<low_mem> options (see L<MEMORY USAGE> below). |
Default is not to use C<low_mem> options (see L<MEMORY USAGE> below). |
79 |
|
|
80 |
|
C<recode> is optional string constisting of character or words pairs that |
81 |
|
should be replaced in input stream. |
82 |
|
|
83 |
|
C<no_progress_bar> disables progress bar output on C<STDOUT> |
84 |
|
|
85 |
This function will also call low-level C<init> if it exists with same |
This function will also call low-level C<init> if it exists with same |
86 |
parametars. |
parametars. |
87 |
|
|
99 |
$module =~ s#::#/#g; |
$module =~ s#::#/#g; |
100 |
$module .= '.pm'; |
$module .= '.pm'; |
101 |
$log->debug("require low-level module $self->{module} from $module"); |
$log->debug("require low-level module $self->{module} from $module"); |
102 |
|
|
103 |
require $module; |
require $module; |
104 |
eval $self->{module} .'->import'; |
#eval $self->{module} .'->import'; |
105 |
|
|
106 |
# check if required subclasses are implemented |
# check if required subclasses are implemented |
107 |
foreach my $subclass (qw/open_db fetch_rec/) { |
foreach my $subclass (qw/open_db fetch_rec init/) { |
108 |
if ( $self->can($subclass) ) { |
my $n = $self->{module} . '::' . $subclass; |
109 |
$log->debug("imported $subclass"); |
if (! defined &{ $n }) { |
110 |
|
my $missing = "missing $subclass in $self->{module}"; |
111 |
|
$self->{$subclass} = sub { $log->logwarn($missing) }; |
112 |
} else { |
} else { |
113 |
$log->warn("missing $subclass in $self->{module}"); |
$self->{$subclass} = \&{ $n }; |
114 |
} |
} |
115 |
} |
} |
116 |
|
|
117 |
if ($self->can('init')) { |
if ($self->{init}) { |
118 |
$log->debug("calling init"); |
$log->debug("calling init"); |
119 |
$self->init(@_); |
$self->{init}->($self, @_); |
120 |
} |
} |
121 |
|
|
122 |
$self->{'code_page'} ||= 'ISO-8859-2'; |
$self->{'code_page'} ||= 'ISO-8859-2'; |
183 |
my $code_page = $arg->{'code_page'} || '852'; |
my $code_page = $arg->{'code_page'} || '852'; |
184 |
|
|
185 |
# store data in object |
# store data in object |
186 |
$self->{'code_page'} = $code_page; |
$self->{'input_code_page'} = $code_page; |
187 |
foreach my $v (qw/path offset limit/) { |
foreach my $v (qw/path offset limit/) { |
188 |
$self->{$v} = $arg->{$v} if ($arg->{$v}); |
$self->{$v} = $arg->{$v} if ($arg->{$v}); |
189 |
} |
} |
191 |
# create Text::Iconv object |
# create Text::Iconv object |
192 |
$self->{iconv} = Text::Iconv->new($code_page,$self->{'code_page'}); |
$self->{iconv} = Text::Iconv->new($code_page,$self->{'code_page'}); |
193 |
|
|
194 |
my ($db, $size) = $self->open_db( |
my $filter_ref; |
195 |
|
|
196 |
|
if ($self->{recode}) { |
197 |
|
my @r = split(/\s/, $self->{recode}); |
198 |
|
if ($#r % 2 != 1) { |
199 |
|
$log->logwarn("recode needs even number of elements (some number of valid pairs)"); |
200 |
|
} else { |
201 |
|
my $recode; |
202 |
|
while (@r) { |
203 |
|
my $from = shift @r; |
204 |
|
my $to = shift @r; |
205 |
|
$recode->{$from} = $to; |
206 |
|
} |
207 |
|
|
208 |
|
my $regex = join '|' => keys %{ $recode }; |
209 |
|
|
210 |
|
$log->debug("using recode regex: $regex"); |
211 |
|
|
212 |
|
$filter_ref = sub { |
213 |
|
my $t = shift; |
214 |
|
$t =~ s/($regex)/$recode->{$1}/g; |
215 |
|
return $t; |
216 |
|
}; |
217 |
|
|
218 |
|
} |
219 |
|
|
220 |
|
} |
221 |
|
|
222 |
|
my ($db, $size) = $self->{open_db}->( $self, |
223 |
path => $arg->{path}, |
path => $arg->{path}, |
224 |
|
filter => $filter_ref, |
225 |
); |
); |
226 |
|
|
227 |
unless ($db) { |
unless (defined($db)) { |
228 |
$log->logwarn("can't open database $arg->{path}, skipping..."); |
$log->logwarn("can't open database $arg->{path}, skipping..."); |
229 |
return; |
return; |
230 |
} |
} |
234 |
return; |
return; |
235 |
} |
} |
236 |
|
|
237 |
my $offset = 1; |
my $from_rec = 1; |
238 |
my $limit = $size; |
my $to_rec = $size; |
239 |
|
|
240 |
if (my $s = $self->{offset}) { |
if (my $s = $self->{offset}) { |
241 |
$log->info("skipping to MFN $s"); |
$log->info("skipping to MFN $s"); |
242 |
$offset = $s; |
$from_rec = $s; |
243 |
} else { |
} else { |
244 |
$self->{offset} = $offset; |
$self->{offset} = $from_rec; |
245 |
} |
} |
246 |
|
|
247 |
if ($self->{limit}) { |
if ($self->{limit}) { |
248 |
$log->info("limiting to ",$self->{limit}," records"); |
$log->debug("limiting to ",$self->{limit}," records"); |
249 |
$limit = $offset + $self->{limit} - 1; |
$to_rec = $from_rec + $self->{limit} - 1; |
250 |
$limit = $size if ($limit > $size); |
$to_rec = $size if ($to_rec > $size); |
251 |
} |
} |
252 |
|
|
253 |
# store size for later |
# store size for later |
254 |
$self->{size} = ($limit - $offset) ? ($limit - $offset + 1) : 0; |
$self->{size} = ($to_rec - $from_rec) ? ($to_rec - $from_rec + 1) : 0; |
255 |
|
|
256 |
$log->info("processing $self->{size} records in $code_page, convert to $self->{code_page}"); |
$log->info("processing $self->{size}/$size records [$from_rec-$to_rec] convert $code_page -> $self->{code_page}"); |
257 |
|
|
258 |
# read database |
# read database |
259 |
for (my $pos = $offset; $pos <= $limit; $mfn++) { |
for (my $pos = $from_rec; $pos <= $to_rec; $pos++) { |
260 |
|
|
261 |
$log->debug("position: $pos\n"); |
$log->debug("position: $pos\n"); |
262 |
|
|
263 |
my $rec = $self->fetch_rec( $db, $pos ); |
my $rec = $self->{fetch_rec}->($self, $db, $pos ); |
264 |
|
|
265 |
|
$log->debug(sub { Dumper($rec) }); |
266 |
|
|
267 |
if (! $rec) { |
if (! $rec) { |
268 |
$log->warn("record $pos empty? skipping..."); |
$log->warn("record $pos empty? skipping..."); |
279 |
# create lookup |
# create lookup |
280 |
$self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); |
$self->{'lookup'}->add( $rec ) if ($rec && $self->{'lookup'}); |
281 |
|
|
282 |
$self->progress_bar($pos,$limit); |
$self->progress_bar($pos,$to_rec) unless ($self->{no_progress_bar}); |
283 |
|
|
284 |
} |
} |
285 |
|
|
287 |
$self->{last_pcnt} = 0; |
$self->{last_pcnt} = 0; |
288 |
|
|
289 |
# store max mfn and return it. |
# store max mfn and return it. |
290 |
$self->{max_pos} = $limit; |
$self->{max_pos} = $to_rec; |
291 |
$log->debug("max_pos: $limit"); |
$log->debug("max_pos: $to_rec"); |
292 |
|
|
293 |
return $size; |
return $size; |
294 |
} |
} |
325 |
return; |
return; |
326 |
} |
} |
327 |
|
|
328 |
$self->progress_bar($mfn,$self->{max_pos}); |
$self->progress_bar($mfn,$self->{max_pos}) unless ($self->{no_progress_bar}); |
329 |
|
|
330 |
my $rec; |
my $rec; |
331 |
|
|