198 |
|
|
199 |
my @v; |
my @v; |
200 |
if ($self->{'lookup_regex'} && $format =~ $self->{'lookup_regex'}) { |
if ($self->{'lookup_regex'} && $format =~ $self->{'lookup_regex'}) { |
201 |
@v = $self->fill_in_to_arr($rec,$format); |
@v = $self->_rec_to_arr($rec,$format,'fill_in'); |
202 |
} else { |
} else { |
203 |
@v = $self->parse_to_arr($rec,$format); |
@v = $self->_rec_to_arr($rec,$format,'parse'); |
204 |
} |
} |
205 |
if (! @v) { |
if (! @v) { |
206 |
$log->debug("$field <",$self->{tag},"> format: $format no values"); |
$log->debug("$field <",$self->{tag},"> format: $format no values"); |
207 |
# next; |
next; |
208 |
} else { |
} else { |
209 |
$log->debug("$field <",$self->{tag},"> format: $format values: ", join(",", @v)); |
$log->debug("$field <",$self->{tag},"> format: $format values: ", join(",", @v)); |
210 |
} |
} |
308 |
sub parse { |
sub parse { |
309 |
my $self = shift; |
my $self = shift; |
310 |
|
|
311 |
my ($rec, $format_utf8, $i) = @_; |
my ($rec, $format_utf8, $i, $rec_size) = @_; |
312 |
|
|
313 |
return if (! $format_utf8); |
return if (! $format_utf8); |
314 |
|
|
357 |
} |
} |
358 |
|
|
359 |
my $found = 0; |
my $found = 0; |
360 |
my $tmp = $self->get_data(\$rec,$3,$4,$r,\$found); |
my $tmp = $self->get_data(\$rec,$3,$4,$r,\$found,$rec_size); |
361 |
|
|
362 |
if ($found) { |
if ($found) { |
363 |
$found_any->{$fld_type} += $found; |
$found_any->{$fld_type} += $found; |
410 |
return $out; |
return $out; |
411 |
} |
} |
412 |
|
|
|
=head2 parse_to_arr |
|
|
|
|
|
Similar to C<parse>, but returns array of all repeatable fields |
|
|
|
|
|
my @arr = $webpac->parse_to_arr($rec,'v250^a'); |
|
|
|
|
|
=cut |
|
|
|
|
|
sub parse_to_arr { |
|
|
my $self = shift; |
|
|
|
|
|
my ($rec, $format_utf8) = @_; |
|
|
|
|
|
my $log = $self->_get_logger(); |
|
|
|
|
|
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
|
|
return if (! $format_utf8); |
|
|
|
|
|
my $i = 0; |
|
|
my @arr; |
|
|
|
|
|
while (my $v = $self->parse($rec,$format_utf8,$i++)) { |
|
|
push @arr, $v; |
|
|
} |
|
|
|
|
|
$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr); |
|
|
|
|
|
return @arr; |
|
|
} |
|
|
|
|
|
|
|
413 |
=head2 fill_in |
=head2 fill_in |
414 |
|
|
415 |
Workhourse of all: takes record from in-memory structure of database and |
Workhourse of all: takes record from in-memory structure of database and |
431 |
This method will automatically decode UTF-8 string to local code page |
This method will automatically decode UTF-8 string to local code page |
432 |
if needed. |
if needed. |
433 |
|
|
434 |
|
There is optional parametar C<$record_size> which can be used to get sizes of |
435 |
|
all C<field^subfield> combinations in this format. |
436 |
|
|
437 |
|
my $text = $webpac->fill_in($rec,'got: v900^a v900^x',0,\$rec_size); |
438 |
|
|
439 |
=cut |
=cut |
440 |
|
|
441 |
sub fill_in { |
sub fill_in { |
443 |
|
|
444 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
445 |
|
|
446 |
my $rec = shift || $log->logconfess("need data record"); |
my ($rec,$format,$i,$rec_size) = @_; |
447 |
my $format = shift || $log->logconfess("need format to parse"); |
|
448 |
|
$log->logconfess("need data record") unless ($rec); |
449 |
|
$log->logconfess("need format to parse") unless($format); |
450 |
|
|
451 |
# iteration (for repeatable fields) |
# iteration (for repeatable fields) |
452 |
my $i = shift || 0; |
$i ||= 0; |
453 |
|
|
454 |
$log->logdie("infitite loop in format $format") if ($i > ($self->{'max_mfn'} || 9999)); |
$log->logdie("infitite loop in format $format") if ($i > ($self->{'max_mfn'} || 9999)); |
455 |
|
|
471 |
# remove filter{...} from beginning |
# remove filter{...} from beginning |
472 |
$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s); |
$filter_name = $1 if ($format =~ s/^filter{([^}]+)}//s); |
473 |
|
|
474 |
# do actual replacement of placeholders |
{ |
475 |
# repeatable fields |
# fix warnings |
476 |
if ($format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found)/ges) { |
no warnings 'uninitialized'; |
477 |
$just_single = 0; |
|
478 |
} |
# do actual replacement of placeholders |
479 |
|
# repeatable fields |
480 |
|
if ($format =~ s/v(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,$i,\$found,$rec_size)/ges) { |
481 |
|
$just_single = 0; |
482 |
|
} |
483 |
|
|
484 |
# non-repeatable fields |
# non-repeatable fields |
485 |
if ($format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found)/ges) { |
if ($format =~ s/s(\d+)(?:\^(\w))?/$self->get_data(\$rec,$1,$2,0,\$found,$rec_size)/ges) { |
486 |
return if ($i > 0 && $just_single); |
return if ($i > 0 && $just_single); |
487 |
|
} |
488 |
} |
} |
489 |
|
|
490 |
if ($found) { |
if ($found) { |
517 |
} |
} |
518 |
|
|
519 |
|
|
520 |
=head2 fill_in_to_arr |
=head2 _rec_to_arr |
521 |
|
|
522 |
Similar to C<fill_in>, but returns array of all repeatable fields. Usable |
Similar to C<parse> and C<fill_in>, but returns array of all repeatable fields. Usable |
523 |
for fields which have lookups, so they shouldn't be parsed but rather |
for fields which have lookups, so they shouldn't be parsed but rather |
524 |
C<fill_id>ed. |
C<paste>d or C<fill_id>ed. Last argument is name of operation: C<paste> or C<fill_in>. |
525 |
|
|
526 |
my @arr = $webpac->fill_in_to_arr($rec,'[v900];;[v250^a]'); |
my @arr = $webpac->fill_in_to_arr($rec,'[v900];;[v250^a]','paste'); |
527 |
|
|
528 |
=cut |
=cut |
529 |
|
|
530 |
sub fill_in_to_arr { |
sub _rec_to_arr { |
531 |
my $self = shift; |
my $self = shift; |
532 |
|
|
533 |
my ($rec, $format_utf8) = @_; |
my ($rec, $format_utf8, $code) = @_; |
534 |
|
|
535 |
my $log = $self->_get_logger(); |
my $log = $self->_get_logger(); |
536 |
|
|
537 |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
$log->logconfess("need HASH as first argument!") if ($rec !~ /HASH/o); |
538 |
return if (! $format_utf8); |
return if (! $format_utf8); |
539 |
|
|
540 |
|
$log->debug("using $code on $format_utf8"); |
541 |
|
|
542 |
my $i = 0; |
my $i = 0; |
543 |
|
my $max = 0; |
544 |
my @arr; |
my @arr; |
545 |
|
my $rec_size = {}; |
546 |
|
|
547 |
while (my $v = $self->fill_in($rec,$format_utf8,$i++)) { |
while ($i <= $max) { |
548 |
push @arr, $v; |
my @v = $self->$code($rec,$format_utf8,$i++,\$rec_size); |
549 |
|
if ($rec_size) { |
550 |
|
foreach my $f (keys %{ $rec_size }) { |
551 |
|
$max = $rec_size->{$f} if ($rec_size->{$f} > $max); |
552 |
|
} |
553 |
|
$log->debug("max set to $max"); |
554 |
|
undef $rec_size; |
555 |
|
} |
556 |
|
if (@v) { |
557 |
|
push @arr, @v; |
558 |
|
} else { |
559 |
|
push @arr, '' if ($max > $i); |
560 |
|
} |
561 |
} |
} |
562 |
|
|
563 |
$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr); |
$log->debug("format '$format_utf8' returned ",--$i," elements: ", sub { join(" | ",@arr) }) if (@arr); |
570 |
|
|
571 |
Returns value from record. |
Returns value from record. |
572 |
|
|
573 |
my $text = $self->get_data(\$rec,$f,$sf,$i,\$found,\$fld_occurances); |
my $text = $self->get_data(\$rec,$f,$sf,$i,\$found,\$rec_size); |
574 |
|
|
575 |
Required arguments are: |
Required arguments are: |
576 |
|
|
590 |
|
|
591 |
=item C<$i> |
=item C<$i> |
592 |
|
|
593 |
index offset for repeatable values ( 0 ... $#occurances ) |
index offset for repeatable values ( 0 ... $rec_size->{'400^a'} ) |
594 |
|
|
595 |
=item C<$found> |
=item C<$found> |
596 |
|
|
597 |
optional variable that will be incremeted if preset |
optional variable that will be incremeted if preset |
598 |
|
|
599 |
=item C<$fld_occurances> |
=item C<$rec_size> |
600 |
|
|
601 |
hash to hold maximum occurances of C<field\tsubfield> combinations |
hash to hold maximum occurances of C<field^subfield> combinations |
602 |
(which can be accessed using keys in same format) |
(which can be accessed using keys in same format) |
603 |
|
|
604 |
=back |
=back |
605 |
|
|
606 |
Returns value or empty string, updates C<$found> and C<fld_occurences> |
Returns value or empty string, updates C<$found> and C<rec_size> |
607 |
if present. |
if present. |
608 |
|
|
609 |
=cut |
=cut |
616 |
return '' unless ($$rec->{$f} && ref($$rec->{$f}) eq 'ARRAY'); |
return '' unless ($$rec->{$f} && ref($$rec->{$f}) eq 'ARRAY'); |
617 |
|
|
618 |
if (defined($$cache)) { |
if (defined($$cache)) { |
619 |
$$cache->{"$f\t$sf"} ||= $$#rec->{$f}; |
$$cache->{ $f . ( $sf ? '^' . $sf : '' ) } ||= scalar @{ $$rec->{$f} }; |
620 |
} |
} |
621 |
|
|
622 |
return '' unless ($$rec->{$f}->[$i]); |
return '' unless ($$rec->{$f}->[$i]); |