2 |
|
|
3 |
use warnings; |
use warnings; |
4 |
use strict; |
use strict; |
5 |
|
use base 'WebPAC::Common'; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
|
|
8 |
=head1 NAME |
=head1 NAME |
11 |
|
|
12 |
=head1 VERSION |
=head1 VERSION |
13 |
|
|
14 |
Version 0.01 |
Version 0.02 |
15 |
|
|
16 |
=cut |
=cut |
17 |
|
|
18 |
our $VERSION = '0.01'; |
our $VERSION = '0.02'; |
19 |
|
|
20 |
=head1 SYNOPSIS |
=head1 SYNOPSIS |
21 |
|
|
79 |
return length($_); |
return length($_); |
80 |
}, ... |
}, ... |
81 |
}, |
}, |
82 |
db => $webpac_db_obj, |
db => $db_obj, |
83 |
lookup_regex => $lookup->regex, |
lookup_regex => $lookup->regex, |
84 |
|
lookup => $lookup_obj, |
85 |
); |
); |
86 |
|
|
87 |
Parametar C<filter> defines user supplied snippets of perl code which can |
Parametar C<filter> defines user supplied snippets of perl code which can |
88 |
be use with C<filter{...}> notation. |
be use with C<filter{...}> notation. |
89 |
|
|
90 |
Recommended parametar C<lookup_regex> is used to enable parsing of lookups |
Recommended parametar C<lookup_regex> is used to enable parsing of lookups |
91 |
in structures. |
in structures. If you pass this parametar, you must also pass C<lookup> |
92 |
|
which is C<WebPAC::Lookup> object. |
93 |
|
|
94 |
=cut |
=cut |
95 |
|
|
98 |
my $self = {@_}; |
my $self = {@_}; |
99 |
bless($self, $class); |
bless($self, $class); |
100 |
|
|
101 |
|
my $r = $self->{'lookup_regex'} ? 1 : 0; |
102 |
|
my $l = $self->{'lookup'} ? 1 : 0; |
103 |
|
|
104 |
|
my $log = $self->_get_logger(); |
105 |
|
|
106 |
|
# those two must be in pair |
107 |
|
if ( ($r & $l) != ($r || $l) ) { |
108 |
|
my $log = $self->_get_logger(); |
109 |
|
$log->logdie("lookup_regex and lookup must be in pair"); |
110 |
|
} |
111 |
|
|
112 |
|
$log->logdie("lookup must be WebPAC::Lookup object") if ($self->{'lookup'} && ! $self->{'lookup'}->isa('WebPAC::Lookup')); |
113 |
|
|
114 |
$self ? return $self : return undef; |
$self ? return $self : return undef; |
115 |
} |
} |
116 |
|
|
122 |
|
|
123 |
This structures are used to produce output. |
This structures are used to produce output. |
124 |
|
|
125 |
my @ds = $webpac->data_structure($rec); |
my $ds = $webpac->data_structure($rec); |
|
|
|
|
B<Note: historical oddity follows> |
|
|
|
|
|
This method will also set C<< $webpac->{'currnet_filename'} >> if there is |
|
|
C<< <filename> >> tag and C<< $webpac->{'headline'} >> if there is |
|
|
C<< <headline> >> tag. |
|
126 |
|
|
127 |
=cut |
=cut |
128 |
|
|
134 |
my $rec = shift; |
my $rec = shift; |
135 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
136 |
|
|
137 |
|
$log->debug("data_structure rec = ", sub { Dumper($rec) }); |
138 |
|
|
139 |
|
$log->logdie("need unique ID (mfn) in field 000 of record ", sub { Dumper($rec) } ) unless (defined($rec->{'000'})); |
140 |
|
|
141 |
|
my $mfn = $rec->{'000'}->[0] || $log->logdie("field 000 isn't array!"); |
142 |
|
|
143 |
my $cache_file; |
my $cache_file; |
144 |
|
|
145 |
if ($self->{'db'}) { |
if ($self->{'db'}) { |
146 |
my @ds = $self->{'db'}->get_ds($rec); |
my $ds = $self->{'db'}->load_ds( $mfn ); |
147 |
return @ds if (@ds); |
$log->debug("load_ds( rec = ", sub { Dumper($rec) }, ") = ", sub { Dumper($ds) }); |
148 |
|
return $ds if ($ds); |
149 |
|
$log->debug("cache miss, creating"); |
150 |
} |
} |
151 |
|
|
152 |
undef $self->{'currnet_filename'}; |
undef $self->{'currnet_filename'}; |
160 |
$self->{tags_by_order} = \@sorted_tags; |
$self->{tags_by_order} = \@sorted_tags; |
161 |
} |
} |
162 |
|
|
163 |
my @ds; |
my $ds; |
164 |
|
|
165 |
$log->debug("tags: ",sub { join(", ",@sorted_tags) }); |
$log->debug("tags: ",sub { join(", ",@sorted_tags) }); |
166 |
|
|
171 |
#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}); |
#print "field $field [",$self->{'tag'},"] = ",Dumper($self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}); |
172 |
|
|
173 |
foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) { |
foreach my $tag (@{$self->{'import_xml'}->{'indexer'}->{$field}->{$self->{'tag'}}}) { |
174 |
my $format = $tag->{'value'} || $tag->{'content'}; |
my $format; |
175 |
|
|
176 |
|
$log->logdie("expected tag HASH and got $tag") unless (ref($tag) eq 'HASH'); |
177 |
|
$format = $tag->{'value'} || $tag->{'content'}; |
178 |
|
|
179 |
$log->debug("format: $format"); |
$log->debug("format: $format"); |
180 |
|
|
195 |
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
@v = map { $self->apply_format($tag->{'format_name'},$tag->{'format_delimiter'},$_) } @v; |
196 |
} |
} |
197 |
|
|
|
if ($field eq 'filename') { |
|
|
$self->{'current_filename'} = join('',@v); |
|
|
$log->debug("filename: ",$self->{'current_filename'}); |
|
|
} elsif ($field eq 'headline') { |
|
|
$self->{'headline'} .= join('',@v); |
|
|
$log->debug("headline: ",$self->{'headline'}); |
|
|
next; # don't return headline in data_structure! |
|
|
} |
|
|
|
|
198 |
# delimiter will join repeatable fields |
# delimiter will join repeatable fields |
199 |
if ($tag->{'delimiter'}) { |
if ($tag->{'delimiter'}) { |
200 |
@v = ( join($tag->{'delimiter'}, @v) ); |
@v = ( join($tag->{'delimiter'}, @v) ); |
201 |
} |
} |
202 |
|
|
203 |
# default types |
# default types |
204 |
my @types = qw(display swish); |
my @types = qw(display search); |
205 |
# override by type attribute |
# override by type attribute |
206 |
@types = ( $tag->{'type'} ) if ($tag->{'type'}); |
@types = ( $tag->{'type'} ) if ($tag->{'type'}); |
207 |
|
|
234 |
|
|
235 |
# TODO: name_sigular, name_plural |
# TODO: name_sigular, name_plural |
236 |
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
my $name = $self->{'import_xml'}->{'indexer'}->{$field}->{'name'}; |
237 |
$row->{'name'} = $name ? $self->_x($name) : $field; |
my $row_name = $name ? $self->_x($name) : $field; |
238 |
|
|
239 |
# post-sort all values in field |
# post-sort all values in field |
240 |
if ($self->{'import_xml'}->{'indexer'}->{$field}->{'sort'}) { |
if ($self->{'import_xml'}->{'indexer'}->{$field}->{'sort'}) { |
241 |
$log->warn("sort at field tag not implemented"); |
$log->warn("sort at field tag not implemented"); |
242 |
} |
} |
243 |
|
|
244 |
push @ds, $row; |
$ds->{$row_name} = $row; |
245 |
|
|
246 |
$log->debug("row $field: ",sub { Dumper($row) }); |
$log->debug("row $field: ",sub { Dumper($row) }); |
247 |
} |
} |
248 |
|
|
249 |
} |
} |
250 |
|
|
251 |
$self->{'db'}->put_gs( |
$self->{'db'}->save_ds( |
252 |
ds => \@ds, |
id => $mfn, |
253 |
current_filename => $self->{'current_filename'}, |
ds => $ds, |
|
headline => $self->{'headline'}, |
|
254 |
) if ($self->{'db'}); |
) if ($self->{'db'}); |
255 |
|
|
256 |
return @ds; |
$log->debug("ds: ", sub { Dumper($ds) }); |
257 |
|
|
258 |
|
$log->logconfess("data structure returned is not array any more!") if wantarray; |
259 |
|
|
260 |
|
return $ds; |
261 |
|
|
262 |
} |
} |
263 |
|
|
453 |
} |
} |
454 |
# do we have lookups? |
# do we have lookups? |
455 |
if ($self->{'lookup'}) { |
if ($self->{'lookup'}) { |
456 |
return $self->lookup($format); |
if ($self->{'lookup'}->can('lookup')) { |
457 |
|
return $self->{'lookup'}->lookup($format); |
458 |
|
} else { |
459 |
|
$log->warn("Have lookup object but can't invoke lookup method"); |
460 |
|
} |
461 |
} else { |
} else { |
462 |
return $format; |
return $format; |
463 |
} |
} |
530 |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
if ($sf && $$rec->{$f}->[$i]->{$sf}) { |
531 |
$$found++ if (defined($$found)); |
$$found++ if (defined($$found)); |
532 |
return $$rec->{$f}->[$i]->{$sf}; |
return $$rec->{$f}->[$i]->{$sf}; |
533 |
} elsif ($$rec->{$f}->[$i]) { |
} elsif (! $sf && $$rec->{$f}->[$i]) { |
534 |
$$found++ if (defined($$found)); |
$$found++ if (defined($$found)); |
535 |
# it still might have subfield, just |
# it still might have subfield, just |
536 |
# not specified, so we'll dump all |
# not specified, so we'll dump all |
543 |
} else { |
} else { |
544 |
return $$rec->{$f}->[$i]; |
return $$rec->{$f}->[$i]; |
545 |
} |
} |
546 |
|
} else { |
547 |
|
return ''; |
548 |
} |
} |
549 |
} else { |
} else { |
550 |
return ''; |
return ''; |
585 |
$log->debug("using format $name [$format] on $data to produce: $out"); |
$log->debug("using format $name [$format] on $data to produce: $out"); |
586 |
|
|
587 |
if ($self->{'lookup_regex'} && $out =~ $self->{'lookup_regex'}) { |
if ($self->{'lookup_regex'} && $out =~ $self->{'lookup_regex'}) { |
588 |
return $self->lookup($out); |
return $self->{'lookup'}->lookup($out); |
589 |
} else { |
} else { |
590 |
return $out; |
return $out; |
591 |
} |
} |