# $Id: Config.pod,v 1.15 2007-09-14 14:41:50 mike Exp $
use strict;
use warnings;
=head1 NAME
Net::Z3950::Simple2ZOOM::Config - configuration file for the Simple2ZOOM gateway
=head1 SYNOPSIS
http://some.url/{user}/pwd={pass}
http://z3950.loc.gov:7090/voyager
get
marc-8
cql
title creator
=head1 DESCRIPTION
The universal Swiss Army Gateway C is configured by a
single file, named on the command-line, and expressed in XML. This
file specifies which back-end databases are supported, how the
back-ends are contacted, what character-sets they provide records in,
and how to map Z39.50 searches to CQL.
The structure of the file is pretty simple.
=head2 Top Level
=over 4
=item
The top-level element is . It contains a single optional
element, any number of elements and a
single optional element. The second of these specifies how
to interpret requests to search in the configured databases; the last
provides query mapping specifications for dynamically specified
databases.
=item
This element contains a URL template, specifying the address of an
HTTP authentication server. The template must include the special
strings C<{user}> and C<{pass}>, which are substituted with the
username and password supplied in the Init request, if any. The
resulting URL is actioned and the result examined: any successful
response (HTTP status 200) indicates that the username/password
combination is acceptable, and that the session can continue; any
other response (e.g. 401 Authorization Required) results in the Init
request being refused with BIB-1 diagnostic 1014 (Init/AC:
Authentication System error).
If the element is omitted from the configuration, no
authentication credentials are required, and any that are provided are
ignored.
(A trivial example of an authentication server script is included in
the Simple2ZOOM distribution, as C.)
=item
The element carries a C attribute specifying the
Z39.50 database name by which is it is known to clients. It contains
several complex elements, and is discussed in more detail below.
=item
Each element, whether contained within a specific
(see below)
or at the top level, consists of a single mandatory
element followed by any number of s. The content of
indicates the type of query that should be sent to the back-end
server, with Simple2ZOOM reposible for translating incoming queries as
required into that format. At present, the only supported value is
C.
=item
Each element carries a C attribute, which is the
numeric value of BIB-1 use attribute to be supported, and optionally
contains a single element which in turn contains the name of
the corresponding CQL index. Type-1 searches against the specified
BIB-1 access point are mapped to CQL searches against the specified
index.
If the is omitted within a , then the generated CQL query
term has no index specified. This can be useful for BIB-1 attributes
such as 1016 (any) and 1035 (anywhere).
=back
=head2 Databases
The element which describes each database contains the
following elements in the specified order.
In general, entries are of two kinds: those connecting
through to a Z39.50 database will have no element, since no
query mapping is necessary to translate an incoming Type-1 query; but
those connecting to an SRU or SRW database will have a
element with set to C and containing information on
how to map from specified BIB-1 use attributes to CQL indexes.
=over 4
=item
Contains the target address of the
back-end database (e.g. C or
C).
=item
Optional. If provided, it must take one of the following values, and
if it is omitted then the value C is assumed:
=over 4
=item C
When queries are received that include references to existing
result-sets, these are translated into result-set references using the
C index. It is an error if the server does not
support this facility.
=item C
References to existing result-sets are rewritten as resubmissions of
the query. This works on all servers, but does not reliably give
precisely correct results if the database is updated between searches.
=item C
Result-set references are used when supported, but resubmissions of
prior queries are used when this facility is unavailable.
=back
=item
This is optional. If provided, it is empty and indicates that the
back-end database does not support named result sets.
=item
There may be any number of these.
Each element carries a C attribute and contains a
corresponding value. These are ZOOM options which are applied to the
connection when it is first created, and can be used to control, for
example, the desired C or C of the records
provided by the back-end. A particularly important option is C,
which may be set to C, C or C to request vanilla SRU,
SRU over POST and SRW respectively.
=item
Optional. Contains the name of the character-set in which the
back-end target supplies records (e.g. C)
=item
Optional. Provides specifications for how to search the database,
exactly like the top-level element described above.
=item
Optional and repeatable: each element indicates special handling for
when records are requested in a particular schema. See below.
=item , , .
Optional. Provides specifications for how to construct records in the
relevant syntaxes when they are requested by clients.
The format is the same in all cases: the specification contains a list
of elements, each of which has an C attribute and
textual content. Records are built by accessing the data addressed by
the specified XPath expressions, and encoding each as an element
addressed as specified by the element content. The interpretation of
the content is different for different record-syntaxes:
=over 4
=item SUTRS
The content is ignored.
=item USMARC
The content indicates a MARC field by a string consisting of the
following parts, in order: a three-digit field number; optionally a
slash followed by the first indicator; optionally another slash
followed by the second indicator; optionally a dollar sign followed by
a subfield tag. In other words, MARC field specifications much match
the regular expression C. It is
impossible to specify the second indicator without the first, but a
subfield may be specified along with zero, one or two indicators.
As usual, a few examples are worth any amount of explanation:
001
260$c
500$a
100/1$a
245/1/0$a
=item GRS-1
The content indicates an address within a GRS-1 record in the form of
one or more consecutive (type,value) pairs, each enclosed in
parentheses. For example, C<(1,14)> would indicate an element of
type 1 (tagSet-M) with value 14 (localControlNumber). A longer path
such as C<(3,admin)(2,6)> indicates an abstract field (tagSet-G
element 6) within an "admin" sub-record.
=back
=back
=head2 Schemas
Each element is empty, but carries the following attributes,
which are used to provide records to Z39.50 clients in MARC formats.
=over 4
=item oid
Mandatory. This is the OID of a Z39.50 record-syntax which is to be
handled by schema mapping. Requests in this database for this
record-syntax are handled as specified.
Example value: C<1.2.840.10003.5.10>
=item sru
Mandatory. This is the URI of an SRU/W schema which is requested from
the SRU or SRW back-end in order to fulfill the request.
Example value: C
=item format
Optional. Indicates which of the MARC variants is in use, so that the
record can be formatted correctly. Defaults to C if omitted.
Example values: C, C, C
=item encoding
Optional. Indicates which character-set to use for the formatted
record. Defaults to C if omitted.
Example values: C,
=back
B that in its current form this schema-mapping only works for
the specific though common combination of Z39.50 front-end, SRU
back-end and MARC record syntax.
=head1 CONFIGURATION FILE SCHEMA
The Simple2ZOOM distribution includes, in the C directory, an XML
schema which can be used to validate configuration files. This schema
is provided in four formats:
=over 4
=item simple2zoom.rnc
Relax-NG compact format: a simple, elegant, terse and wholly
comprehensible XML constraint language that you don't even need to
learn in order to understand. This is the master version: the others
are automatically generated from it.
=item simple2zoom.rng
Relax-NG XML format: the world seems to have this zany fetish that
everything must be specified in XML, so Relax-NG has an XML syntax
that corresponds trivially with the much nicer compact syntax. The
principle value of this is that C understands it.
=item simple2zoom.dtd
An old-fashioned DTD (document type definition).
=item simple2zoom.xsd
If you must.
=back
Use whichever you like. For example,
xmllint --relaxng simple2zoom.rng --noout test.xml
xmllint --dtdvalid simple2zoom.dtd --noout test.xml
xmllint --schema simple2zoom.xsd --noout test.xml
=head1 SEE ALSO
The C program.
The C module.
=head1 AUTHOR
Mike Taylor Emike@indexdata.comE
=head1 COPYRIGHT AND LICENCE
Copyright (C) 2007 by Index Data.
This library is free software; you can redistribute it and/or modify
it under the same terms as Perl itself, either Perl version 5.8.8 or,
at your option, any later version of Perl 5 you may have available.
=cut
1;