[EP-tech] Re: OAI-PMH

[Florian Heß]

Am 04.07.2014 15:34, schrieb Ian Stuart:
> But have a good look at what the
> http://<your.server.address/cgi/oai2?verb=ListSets returns (and there's
> a reason for those bloody horrible SetNames)

Hi Ian,

For special needs besides of what can be done with the advanced search, 
EPrints should support partially dynamic but human-readable set names as 
defined in the configuration. We have OAI clients that *expect* set 
names composed in a certain way, e.g. containing literal field values as 
in "ddc:300" or "doc-type:article".

We must use a hack here to include a list of kind-of-automatically 
pregenerated custom sets into the configuration. I would humbly prefer 
some better (e.g. out of the box) method to achieve that, admittedly:


# in cfg.d/oai.pl:
use UBHD::EPrints::DINISets2010 qw(cfg_setspecs);

$oai->{custom_sets} = [
         cfg_setspecs(),
       # ...
];

# in a .pm file residing in an %INC directory:

package UBHD::EPrints::DINISets2010;
use base 'Exporter';

our @EXPORT_OK = qw( sets_for_eprint cfg_setspecs );

my %fields = (
    'ddc' => {
        spec => sub {{ filters => [{ meta_fields => [ "subjects" ], 
value => shift }] }},
        eprint => sub { my $ep = shift; return map { "ddc:$_" } 
@{$ep->value("subjects")}; }
    },
    'doc-type' => {
        spec => sub {{ filters => [{ meta_fields => [ "type" ], value => 
shift }] }},
        eprint => sub { my $ep = shift; return 
"doc-type:".$ep->value("type") },
    },
);

my %sets;

while ( defined($_ = <DATA>) ) {
     chomp; ($_) = split /\s*#/; next if !$_;
     my ($setSpec, $type, $value, $name) = m{ \A (([\w-]+):(\S+)) \s 
(.+) \z }xms
         or die "Odd format in line!";
     my $spec = $sets{$setSpec} = $fields{$type}{spec}->($value);
     $spec->{spec} = $setSpec;
     $spec->{name} = $name;
}
close DATA;

sub cfg_setspecs { values %sets }

sub sets_for_eprint {
     my ($eprint) = @_;
     return map { $_->{eprint}->($eprint) } values %fields;
}

1;

__DATA__
ddc:000 Generalities, Science
ddc:004 Data processing Computer science
ddc:010 Bibliography
...
doc-type:preprint Preprint
doc-type:workingPaper WorkingPaper
doc-type:article Article
doc-type:PeriodicalPart PeriodicalPart
doc-type:Periodical Periodical
doc-type:book Book
...


Florian

-- 
UB Heidelberg (Altstadt)
Plöck 107-109, 69117 HD
Abt. Informationstechnik
http://www.ub.uni-heidelberg.de/