3 |
use warnings; |
use warnings; |
4 |
use strict; |
use strict; |
5 |
|
|
6 |
use blib; |
use lib 'lib'; |
7 |
|
|
8 |
use WebPAC::Common; |
use WebPAC::Common; |
9 |
use base qw/WebPAC::Common/; |
use base qw/WebPAC::Common/; |
10 |
use Data::Dump qw/dump/; |
use Data::Dump qw/dump/; |
11 |
use Encode qw/from_to/; |
use Encode qw/decode from_to/; |
12 |
|
use YAML; |
13 |
|
|
14 |
=head1 NAME |
=head1 NAME |
15 |
|
|
16 |
WebPAC::Input - read different file formats into WebPAC |
WebPAC::Input - read different file formats into WebPAC |
17 |
|
|
|
=head1 VERSION |
|
|
|
|
|
Version 0.18 |
|
|
|
|
18 |
=cut |
=cut |
19 |
|
|
20 |
our $VERSION = '0.18'; |
our $VERSION = '0.19'; |
21 |
|
|
22 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
23 |
|
|
58 |
|
|
59 |
my $db = new WebPAC::Input( |
my $db = new WebPAC::Input( |
60 |
module => 'WebPAC::Input::MARC', |
module => 'WebPAC::Input::MARC', |
|
encoding => 'ISO-8859-2', |
|
61 |
recode => 'char pairs', |
recode => 'char pairs', |
62 |
no_progress_bar => 1, |
no_progress_bar => 1, |
63 |
|
input_config => { |
64 |
|
mapping => [ 'foo', 'bar', 'baz' ], |
65 |
|
}, |
66 |
); |
); |
67 |
|
|
68 |
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 |
69 |
L<WebPAC::Input::MARC>. |
L<WebPAC::Input::MARC>. |
70 |
|
|
|
Optional parametar C<encoding> specify application code page (which will be |
|
|
used internally). This should probably be your terminal encoding, and by |
|
|
default, it C<ISO-8859-2>. |
|
|
|
|
71 |
C<recode> is optional string constisting of character or words pairs that |
C<recode> is optional string constisting of character or words pairs that |
72 |
should be replaced in input stream. |
should be replaced in input stream. |
73 |
|
|
85 |
|
|
86 |
my $log = $self->_get_logger; |
my $log = $self->_get_logger; |
87 |
|
|
88 |
$log->logconfess("code_page argument is not suppored any more. change it to encoding") if ($self->{lookup}); |
$log->logconfess("code_page argument is not suppored any more.") if $self->{code_page}; |
89 |
$log->logconfess("lookup argument is not suppored any more. rewrite call to lookup_ref") if ($self->{lookup}); |
$log->logconfess("encoding argument is not suppored any more.") if $self->{encoding}; |
90 |
$log->logconfess("low_mem argument is not suppored any more. rewrite it to load_row and save_row") if ($self->{low_mem}); |
$log->logconfess("lookup argument is not suppored any more. rewrite call to lookup_ref") if $self->{lookup}; |
91 |
|
$log->logconfess("low_mem argument is not suppored any more. rewrite it to load_row and save_row") if $self->{low_mem}; |
92 |
|
|
93 |
$log->logconfess("specify low-level file format module") unless ($self->{module}); |
$log->logconfess("specify low-level file format module") unless ($self->{module}); |
94 |
my $module_path = $self->{module}; |
my $module_path = $self->{module}; |
98 |
|
|
99 |
require $module_path; |
require $module_path; |
100 |
|
|
|
$self->{'encoding'} ||= 'ISO-8859-2'; |
|
|
|
|
101 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
102 |
} |
} |
103 |
|
|
109 |
|
|
110 |
$input->open( |
$input->open( |
111 |
path => '/path/to/database/file', |
path => '/path/to/database/file', |
112 |
code_page => 'cp852', |
input_encoding => 'cp852', |
113 |
|
strict_encoding => 0, |
114 |
limit => 500, |
limit => 500, |
115 |
offset => 6000, |
offset => 6000, |
116 |
stats => 1, |
stats => 1, |
135 |
|
|
136 |
); |
); |
137 |
|
|
138 |
By default, C<code_page> is assumed to be C<cp852>. |
By default, C<input_encoding> is assumed to be C<cp852>. |
139 |
|
|
140 |
C<offset> is optional parametar to position at some offset before reading from database. |
C<offset> is optional parametar to position at some offset before reading from database. |
141 |
|
|
156 |
C<save_row> and C<load_row> are low-level implementation of store engine. Calling convention |
C<save_row> and C<load_row> are low-level implementation of store engine. Calling convention |
157 |
is documented in example above. |
is documented in example above. |
158 |
|
|
159 |
|
C<strict_encoding> should really default to 1, but it doesn't for now. |
160 |
|
|
161 |
Returns size of database, regardless of C<offset> and C<limit> |
Returns size of database, regardless of C<offset> and C<limit> |
162 |
parametars, see also C<size>. |
parametars, see also C<size>. |
163 |
|
|
168 |
my $arg = {@_}; |
my $arg = {@_}; |
169 |
|
|
170 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
171 |
|
$log->debug( "arguments: ",dump( $arg )); |
172 |
|
|
173 |
|
$log->logconfess("encoding argument is not suppored any more.") if $self->{encoding}; |
174 |
|
$log->logconfess("code_page argument is not suppored any more.") if $self->{code_page}; |
175 |
$log->logconfess("lookup argument is not suppored any more. rewrite call to lookup_coderef") if ($arg->{lookup}); |
$log->logconfess("lookup argument is not suppored any more. rewrite call to lookup_coderef") if ($arg->{lookup}); |
176 |
$log->logconfess("lookup_coderef must be CODE, not ",ref($arg->{lookup_coderef})) |
$log->logconfess("lookup_coderef must be CODE, not ",ref($arg->{lookup_coderef})) |
177 |
if ($arg->{lookup_coderef} && ref($arg->{lookup_coderef}) ne 'CODE'); |
if ($arg->{lookup_coderef} && ref($arg->{lookup_coderef}) ne 'CODE'); |
179 |
$log->debug( $arg->{lookup_coderef} ? '' : 'not ', "using lookup_coderef"); |
$log->debug( $arg->{lookup_coderef} ? '' : 'not ', "using lookup_coderef"); |
180 |
|
|
181 |
$log->logcroak("need path") if (! $arg->{'path'}); |
$log->logcroak("need path") if (! $arg->{'path'}); |
182 |
my $code_page = $arg->{'code_page'} || 'cp852'; |
my $input_encoding = $arg->{'input_encoding'} || $self->{'input_encoding'} || 'cp852'; |
183 |
|
|
184 |
# store data in object |
# store data in object |
|
$self->{'input_code_page'} = $code_page; |
|
185 |
foreach my $v (qw/path offset limit/) { |
foreach my $v (qw/path offset limit/) { |
186 |
$self->{$v} = $arg->{$v} if ($arg->{$v}); |
$self->{$v} = $arg->{$v} if defined $arg->{$v}; |
187 |
} |
} |
188 |
|
|
189 |
if ($arg->{load_row} || $arg->{save_row}) { |
if ($arg->{load_row} || $arg->{save_row}) { |
230 |
|
|
231 |
my $class = $self->{module} || $log->logconfess("can't get low-level module name!"); |
my $class = $self->{module} || $log->logconfess("can't get low-level module name!"); |
232 |
|
|
233 |
|
$arg->{$_} = $self->{$_} foreach qw(offset limit); |
234 |
|
|
235 |
my $ll_db = $class->new( |
my $ll_db = $class->new( |
236 |
path => $arg->{path}, |
path => $arg->{path}, |
237 |
|
input_config => $arg->{input_config} || $self->{input_config}, |
238 |
# filter => sub { |
# filter => sub { |
239 |
# my ($l,$f_nr) = @_; |
# my ($l,$f_nr) = @_; |
240 |
# return unless defined($l); |
# return unless defined($l); |
241 |
# from_to($l, $code_page, $self->{'encoding'}); |
# $l = decode($input_encoding, $l); |
242 |
# $l =~ s/($recode_regex)/$recode_map->{$1}/g if ($recode_regex && $recode_map); |
# $l =~ s/($recode_regex)/$recode_map->{$1}/g if ($recode_regex && $recode_map); |
243 |
# return $l; |
# return $l; |
244 |
# }, |
# }, |
245 |
%{ $arg }, |
%{ $arg }, |
246 |
); |
); |
247 |
|
|
248 |
|
# save for dump and input_module |
249 |
|
$self->{ll_db} = $ll_db; |
250 |
|
|
251 |
unless (defined($ll_db)) { |
unless (defined($ll_db)) { |
252 |
$log->logwarn("can't open database $arg->{path}, skipping..."); |
$log->logwarn("can't open database $arg->{path}, skipping..."); |
253 |
return; |
return; |
279 |
# store size for later |
# store size for later |
280 |
$self->{size} = ($to_rec - $from_rec) ? ($to_rec - $from_rec + 1) : 0; |
$self->{size} = ($to_rec - $from_rec) ? ($to_rec - $from_rec + 1) : 0; |
281 |
|
|
282 |
$log->info("processing $self->{size}/$size records [$from_rec-$to_rec] convert $code_page -> $self->{encoding}", $self->{stats} ? ' [stats]' : ''); |
my $strict_encoding = $arg->{strict_encoding} || $self->{strict_encoding}; ## FIXME should be 1 really |
283 |
|
|
284 |
|
$log->info("processing $self->{size}/$size records [$from_rec-$to_rec]", |
285 |
|
" encoding $input_encoding ", $strict_encoding ? ' [strict]' : '', |
286 |
|
$self->{stats} ? ' [stats]' : '', |
287 |
|
); |
288 |
|
|
289 |
# read database |
# read database |
290 |
for (my $pos = $from_rec; $pos <= $to_rec; $pos++) { |
for (my $pos = $from_rec; $pos <= $to_rec; $pos++) { |
302 |
$log->debug("-=> $f_nr ## $l"); |
$log->debug("-=> $f_nr ## $l"); |
303 |
|
|
304 |
# codepage conversion and recode_regex |
# codepage conversion and recode_regex |
305 |
from_to($l, $code_page, $self->{'encoding'}); |
$l = decode($input_encoding, $l, 1); |
306 |
$l =~ s/($recode_regex)/$recode_map->{$1}/g if ($recode_regex && $recode_map); |
$l =~ s/($recode_regex)/$recode_map->{$1}/g if ($recode_regex && $recode_map); |
307 |
|
|
308 |
# apply regexps |
# apply regexps |
321 |
warn "*** $d\n" if ($debug); |
warn "*** $d\n" if ($debug); |
322 |
|
|
323 |
} |
} |
324 |
$log->error("error applying regex: $r") if ($@); |
$log->error("error applying regex: ",dump($r), $@) if $@; |
325 |
} |
} |
326 |
} |
} |
327 |
|
|
359 |
foreach my $fld (keys %{ $rec }) { |
foreach my $fld (keys %{ $rec }) { |
360 |
$self->{_stats}->{fld}->{ $fld }++; |
$self->{_stats}->{fld}->{ $fld }++; |
361 |
|
|
362 |
$log->logdie("invalid record fild $fld, not ARRAY") |
#$log->logdie("invalid record fild $fld, not ARRAY") |
363 |
unless (ref($rec->{ $fld }) eq 'ARRAY'); |
next unless (ref($rec->{ $fld }) eq 'ARRAY'); |
364 |
|
|
365 |
foreach my $row (@{ $rec->{$fld} }) { |
foreach my $row (@{ $rec->{$fld} }) { |
366 |
|
|
391 |
$self->{max_pos} = $to_rec; |
$self->{max_pos} = $to_rec; |
392 |
$log->debug("max_pos: $to_rec"); |
$log->debug("max_pos: $to_rec"); |
393 |
|
|
|
# save for dump |
|
|
$self->{ll_db} = $ll_db; |
|
|
|
|
394 |
return $size; |
return $size; |
395 |
} |
} |
396 |
|
|
397 |
|
sub input_module { $_[0]->{ll_db} } |
398 |
|
|
399 |
=head2 fetch |
=head2 fetch |
400 |
|
|
401 |
Fetch next record from database. It will also displays progress bar. |
Fetch next record from database. It will also displays progress bar. |
536 |
my $o = sprintf("%4s %d ~", $f, $v); |
my $o = sprintf("%4s %d ~", $f, $v); |
537 |
|
|
538 |
if (defined($s->{sf}->{$f})) { |
if (defined($s->{sf}->{$f})) { |
539 |
|
my @subfields = keys %{ $s->{sf}->{$f} }; |
540 |
map { |
map { |
541 |
$o .= sprintf(" %s:%d%s", $_, |
$o .= sprintf(" %s:%d%s", $_, |
542 |
$s->{sf}->{$f}->{$_}->{count}, |
$s->{sf}->{$f}->{$_}->{count}, |
543 |
$s->{sf}->{$f}->{$_}->{repeatable} ? '*' : '', |
$s->{sf}->{$f}->{$_}->{repeatable} ? '*' : '', |
544 |
); |
); |
545 |
} sort keys %{ $s->{sf}->{$f} }; |
} ( |
546 |
|
# first indicators and other special subfields |
547 |
|
sort( grep { length($_) > 1 } @subfields ), |
548 |
|
# then subfileds (single char) |
549 |
|
sort( grep { length($_) == 1 } @subfields ), |
550 |
|
); |
551 |
} |
} |
552 |
|
|
553 |
if (my $v_r = $s->{repeatable}->{$f}) { |
if (my $v_r = $s->{repeatable}->{$f}) { |
555 |
} |
} |
556 |
|
|
557 |
$o; |
$o; |
558 |
} sort { $a <=> $b } keys %{ $s->{fld} } |
} sort { |
559 |
|
if ( $a =~ m/^\d+$/ && $b =~ m/^\d+$/ ) { |
560 |
|
$a <=> $b |
561 |
|
} else { |
562 |
|
$a cmp $b |
563 |
|
} |
564 |
|
} keys %{ $s->{fld} } |
565 |
); |
); |
566 |
|
|
567 |
$log->debug( sub { dump($s) } ); |
$log->debug( sub { dump($s) } ); |
568 |
|
|
569 |
|
my $path = 'var/stats.yml'; |
570 |
|
YAML::DumpFile( $path, $s ); |
571 |
|
$log->info( 'created ', $path, ' with ', -s $path, ' bytes' ); |
572 |
|
|
573 |
return $out; |
return $out; |
574 |
} |
} |
575 |
|
|
584 |
|
|
585 |
return unless $self->{ll_db}; |
return unless $self->{ll_db}; |
586 |
|
|
587 |
if ($self->{ll_db}->can('dump_rec')) { |
if ($self->{ll_db}->can('dump_ascii')) { |
588 |
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
return $self->{ll_db}->dump_ascii( $self->{pos} ); |
589 |
} else { |
} else { |
590 |
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
return dump( $self->{ll_db}->fetch_rec( $self->{pos} ) ); |
728 |
line => $., |
line => $., |
729 |
}; |
}; |
730 |
$log->debug("regex: $regex"); |
$log->debug("regex: $regex"); |
731 |
|
} else { |
732 |
|
die "can't parse: $_"; |
733 |
} |
} |
734 |
} |
} |
735 |
|
|