my ($record, $id) = @_;
my %marc = (); $marc{isbns} = [];
+ # reset misc score
+ $conf->{misc_score} = 999;
+
# id, stringified
$marc{id} = $id->as_string($conf->{subfield});
Assign a score to the record based on various criteria.
-Score is constructed by pushing elements onto a list. At the end of
-the routine, the list is flattened into a string via join();
+Score is constructed by pushing elements onto a list, via a dispatch
+table. This allows order of fingerprints in the output file to be
+varied.
=cut
sub score_marc {
my ($marc, $record) = @_;
my @score = ();
- my $chunk;
-
- # Is this an OCLC record?
- if ($conf->{scores}{oclc}) {
- if ($record->field('008')) {
- $chunk = $record->field('008')->as_string();
- push @score, ( $chunk =~ /^o/i ? 1 : 0 );
- } else {
- push @score, 0;
- }
- }
-
- # does 040a contain "dlc"?
- if ($conf->{scores}{dlc}) {
- if ($record->field('040') and $record->field('040')->subfield('a')) {
- $chunk = $record->field('040')->subfield('a');
- push @score, ( $chunk =~ /dlc/i ? 1 : 0 );
- } else {
- push @score, 0;
- }
- }
-
- # number of 650 datafields
- # zero-padded to 4 digits with printf
- if ($conf->{scores}{num_650}) {
- if ($record->field('650')) {
- my @tags = $record->field('650');
- push @score, scalar @tags;
- } else {
- push @score, 0;
- }
- }
-
- # number of tags in total
- # zero-padded to 4 digits with printf
- if ($conf->{scores}{num_tags}) {
- my @tags = $record->fields;
- push @score, scalar @tags;
- }
-
- # encoding level
- if ($conf->{scores}{enc_lvl}) {
- my $enc = substr($record->leader, 17, 1);
+ my $json = '{';
+
+ my %scores_code = (
+ oclc => sub { return $marc->{oclc}[0] ? 1 : 0 },
+ dlc => sub {
+ if ($record->field('040') and $record->field('040')->subfield('a'))
+ { return scalar($record->subfield( '040', 'a')) =~ /dlc/io ? 1 : 0 }
+ else { return 0 }
+ },
+ num_650 => sub {
+ if ($record->field('650')) {
+ # can't say "scalar $record->field('650')"; MARC::Record
+ # behaves differently in list/scalar contexts
+ my @tags = $record->field('650');
+ return sprintf("%04d", scalar @tags)
+ } else { return '0000' }
+ },
+ num_tags=> sub { return sprintf( '%04d', scalar( $record->fields ) ) },
+ enc_lvl => sub {
+ my $enc = substr($record->leader, 17, 1) || 'u';
my %levels = ( ' ' => 9, 1 => 8, 2 => 7, 3 => 6, 4 => 5, 5 => 4,
6 => 3, 7 => 2, 8 => 1, 'u' => 0, 'z' => 0 );
- if (defined $enc and $levels{$enc})
- { push @score, $levels{$enc} }
- else
- { push @score, 0 }
+ return $levels{$enc} || 0;
+ }
+ );
+
+ for ( @{ $conf->{dyn_scores} } ) {
+ push @score, $scores_code{$_}->($marc, $record);
+ $json .= $_ . ':' . $score[-1] . ',';
}
- # put score in marc hash
- $marc->{score} = join('', '{oclc:', $score[0], ',dlc:', $score[1],
- ',num_650:', $score[2], ',num_tags:', $score[3],
- ',enc_lvl:', $score[4], '}');
+ # add misc score
+ $json .= 'misc:' . $conf->{misc_score};
+ $json .= '}';
+
+ my $compact = join('', @score, $conf->{misc_score});
+ $marc->{score} = "$compact\t$json";
}
=head2 dump_fingerprints
binmode(STDIN, ':utf8');
my $rc = GetOptions( $c,
- 'incoming',
- 'incumbent',
'exception|x=s',
- 'marctype|m=s',
'output|o=s',
- 'runtype|r=s',
+ 'prefix|p=s',
+ 'marctype|m=s',
'subfield|s=s',
'tag|t=s',
'fingerprints=s',
$c->{fingerprints} = {oclc => 1, isbn => 1, edition => 1, issn => 1,
lccn => 1, accomp => 1, authpub => 1};
}
+
# check scores list for validity
if ($c->{scores}) {
my %scores = ();
die "Invalid score mode '$_'\n" unless $valid_scores{$_};
$scores{$_} = 1;
}
+ $c->{dyn_scores} = [split /,/, $c->{scores}];
$c->{scores} = \%scores;
} else {
$c->{scores} = {oclc => 1, dlc => 1, num_650 => 1,
num_tags => 1, enc_lvl => 1};
+ $c->{dyn_scores} = [ qw/oclc dlc num_650 num_tags enc_lvl/ ];
}
- # set defaults if told to do so
- if ($c->{incoming}) {
- $c->{tag} = 903 unless defined $c->{tag};
- $c->{subfield} = 'a' unless defined $c->{subfield};
- $c->{marctype} = 'XML' unless defined $c->{marctype};
- $c->{output} = 'incoming.fp' unless defined $c->{output};
- $c->{exception} = 'incoming.ex' unless defined $c->{exception};
- } elsif ($c->{incumbent}) {
- $c->{tag} = 901 unless defined $c->{tag};
- $c->{subfield} = 'c' unless defined $c->{subfield};
- $c->{marctype} = 'XML' unless defined $c->{marctype};
- $c->{output} = 'incumbent.fp' unless defined $c->{output};
- $c->{exception} = 'incumbent.ex' unless defined $c->{exception};
+ # set defaults
+ $c->{tag} = 903 unless defined $c->{tag};
+ $c->{subfield} = 'a' unless defined $c->{subfield};
+ $c->{marctype} = 'XML' unless defined $c->{marctype};
+ if ($c->{prefix}) {
+ $c->{output} = join('.',$c->{prefix},'fp');
+ $c->{exception} = join('.',$c->{prefix},'fp','ex');
}
my @keys = keys %{$c};
print <<HELP;
Usage is: $0 [REQUIRED ARGS] [OPTIONS] <filelist>
Req'd Arguments
- --tag=N -t Which tag to use
- --subfield=X -s Which subfield to use
- --output=<file> -o Output filename
- --exceptions=<file> -x Exception report filename
+ --output=<FILE> -o Output filename
+ --exceptions=<FILE> -x Exception report filename
+ or
+ --prefix=<PREFIX>> -p Shared prefix for output/exception files. Will
+ produce PREFIX.fp and PREFIX.fp.ex
Options
- --incoming '-t 903 -s a -o incoming.fp -x incoming.ex'
- --incumbent '-t 901 -s c -o incumbent.fp -x incumbent.ex'
+ --tag=N -t Which tag to use (default 903)
+ --subfield=X -s Which subfield to use (default 'a')
+ --quiet -q Don't write status messages to STDOUT
--fingerprints=LIST Fingerprints to generate, comma separated
Default: oclc,isbn,edition,issn,lccn,accomp,authpub
Others: baseline
+
--scores=LIST Scores to calculate, comma separated
Default: oclc,dlc,num_650,num_tags,enc_level
- --quiet -q Don't write status messages to STDOUT
+
+ --marctype=TYPE Defaults to 'XML'
HELP
exit 1;
}
projects / migration-tools.git / blobdiff