1 <!-- $Id: book.xml,v 1.24 2006-04-27 16:16:28 mike Exp $ -->
3 <title>Metaproxy - User's Guide and Reference</title>
5 <firstname>Mike</firstname><surname>Taylor</surname>
8 <firstname>Adam</firstname><surname>Dickmeiss</surname>
12 <holder>Index Data ApS</holder>
16 Metaproxy is a universal router, proxy and encapsulated
17 metasearcher for information retrieval protocols. It accepts,
18 processes, interprets and redirects requests from IR clients using
19 standard protocols such as
20 <ulink url="&url.z39.50;">ANSI/NISO Z39.50</ulink>
21 (and in the future <ulink url="&url.sru;">SRU</ulink>
22 and <ulink url="&url.srw;">SRW</ulink>), as
23 well as functioning as a limited
24 <ulink url="&url.http;">HTTP</ulink> server.
25 Metaproxy is configured by an XML file which
26 specifies how the software should function in terms of routes that
27 the request packets can take through the proxy, each step on a
28 route being an instantiation of a filter. Filters come in many
29 types, one for each operation: accepting Z39.50 packets, logging,
30 query transformation, multiplexing, etc. Further filter-types can
31 be added as loadable modules to extend Metaproxy functionality,
35 The terms under which Metaproxy will be distributed have yet to be
36 established, but it will not necessarily be open source; so users
37 should not at this stage redistribute the code without explicit
38 written permission from the copyright holders, Index Data ApS.
43 <imagedata fileref="common/id.png" format="PNG"/>
46 <imagedata fileref="common/id.eps" format="EPS"/>
53 <chapter id="introduction">
54 <title>Introduction</title>
58 <ulink url="&url.metaproxy;">Metaproxy</ulink>
59 is a standalone program that acts as a universal router, proxy and
60 encapsulated metasearcher for information retrieval protocols such
61 as <ulink url="&url.z39.50;">Z39.50</ulink>, and in the future
62 <ulink url="&url.sru;">SRU</ulink> and <ulink url="&url.srw;">SRW</ulink>.
63 To clients, it acts as a server of these protocols: it can be searched,
64 records can be retrieved from it, etc.
65 To servers, it acts as a client: it searches in them,
66 retrieves records from them, etc. it satisfies its clients'
67 requests by transforming them, multiplexing them, forwarding them
68 on to zero or more servers, merging the results, transforming
69 them, and delivering them back to the client. In addition, it
70 acts as a simple <ulink url="&url.http;">HTTP</ulink> server; support
71 for further protocols can be added in a modular fashion, through the
72 creation of new filters.
77 Fish, bananas, cold pyjamas,
78 Mutton, beef and trout!
79 - attributed to Cole Porter.
82 Metaproxy is a more capable alternative to
83 <ulink url="&url.yazproxy;">YAZ Proxy</ulink>,
84 being more powerful, flexible, configurable and extensible. Among
85 its many advantages over the older, more pedestrian work are
86 support for multiplexing (encapsulated metasearching), routing by
87 database name, authentication and authorisation and serving local
88 files via HTTP. Equally significant, its modular architecture
89 facilitites the creation of pluggable modules implementing further
93 This manual will briefly describe Metaproxy's licensing situation
94 before giving an overview of its architecture, then discussing the
95 key concept of a filter in some depth and giving an overview of
96 the various filter types, then discussing the configuration file
97 format. After this come several optional chapters which may be
98 freely skipped: a detailed discussion of virtual databases and
99 multi-database searching, some notes on writing extensions
100 (additional filter types) and a high-level description of the
101 source code. Finally comes the reference guide, which contains
102 instructions for invoking the <command>metaproxy</command>
103 program, and detailed information on each type of filter,
110 <chapter id="licence">
111 <title>The Metaproxy Licence</title>
113 <emphasis role="strong">
114 No decision has yet been made on the terms under which
115 Metaproxy will be distributed.
117 It is possible that, unlike
118 other Index Data products, metaproxy may not be released under a
119 free-software licence such as the GNU GPL. Until a decision is
120 made and a public statement made, then, and unless it has been
121 delivered to you other specific terms, please treat Metaproxy as
122 though it were proprietary software.
123 The code should not be redistributed without explicit
124 written permission from the copyright holders, Index Data ApS.
128 <chapter id="installation">
129 <title>Installation</title>
131 Metaproxy depends on the following tools/libraries:
133 <varlistentry><term><ulink url="&url.yazplusplus;">YAZ++</ulink></term>
136 This is a C++ library based on <ulink url="&url.yaz;">YAZ</ulink>.
140 <varlistentry><term><ulink url="&url.libxslt;">Libxslt</ulink></term>
142 <para>This is an XSLT processor - based on
143 <ulink url="&url.libxml2;">Libxml2</ulink>. Both Libxml2 and
144 Libxslt must be installed with the development components
145 (header files, etc.) as well as the run-time libraries.
149 <varlistentry><term><ulink url="&url.boost;">Boost</ulink></term>
152 The popular C++ library. Initial versions of Metaproxy
153 was built with 1.33.0. Version 1.33.1 works too.
160 In order to compile Metaproxy a modern C++ compiler is
161 required. Boost, in particular, requires the C++ compiler
162 to facilitate the newest features. Refer to Boost
163 <ulink url="&url.boost.compilers.status;">Compiler Status</ulink>
164 for more information.
167 We have succesfully used Metaproxy with Boost using the compilers
168 <ulink url="&url.gcc;">GCC</ulink> version 4.0 and
169 <ulink url="&url.vstudio;">Microsoft Visual Studio</ulink> 2003/2005.
172 <section id="installation.unix">
173 <title>Installation on Unix (from Source)</title>
175 Here is a quick step-by-step guide on how to compile all the
176 tools that Metaproxy uses. Only few systems have none of the required
177 tools binary packages. If, for example, Libxml2/libxslt are already
178 installed as development packages use those (and omit compilation).
185 gunzip -c libxml2-version.tar.gz|tar xf -
193 gunzip -c libxslt-version.tar.gz|tar xf -
204 gunzip -c yaz-version.tar.gz|tar xf -
212 gunzip -c yazpp-version.tar.gz|tar xf -
223 gunzip -c boost-version.tar.gz|tar xf -
234 gunzip -c metaproxy-version.tar.gz|tar xf -
243 <section id="installation.debian">
244 <title>Installation on Debian</title>
250 <section id="installation.windows">
251 <title>Installation on Windows</title>
253 Compilation of Metaproxy can be done using
254 Microsoft <ulink url="&url.vstudio;">Visual Studio</ulink>.
255 We know Version 2003 works. We expect Version 2005 to
258 <section id="installation.windows.boost">
261 Get Boost from its <ulink url="&url.boost;">home page</ulink>.
262 You also need Boost Jam (an alternative to make).
263 That's also available from this
264 home page. The files download are called something like:
265 <literal>boost_1_33-1.exe</literal>
267 <literal>boost-jam-3.1.12-1-ntx86.zip</literal>.
268 Unpack Boost Jam first. Put <literal>bjam.exe</literal>
269 in your system path. Make a command prompt and ensure
270 it can be found automatically. If not check the PATH.
271 The Boost .exe is a self-extracting exe with
272 complete source for Boost. Compile that source with
273 Boost Jam (An alternative to Make).
274 The compilation takes a while.
275 By default, the Boost build process puts the resulting
276 libraries + header files in
277 <literal>\boost\lib</literal>, <literal>\boost\include</literal>.
280 For more informatation about installing Boost refer to the
281 <ulink url="&url.boost.getting.started;">getting started</ulink>
286 <section id="installation.windows.libxslt">
287 <title>Libxslt</title>
289 <ulink url="&url.libxslt;">Libxslt</ulink> can be downloaded
291 <ulink url="&url.libxml2.download.win32;">here</ulink>.
294 Libxslt has other dependencies, but thes can all be downloaded
295 from the same site. Get the following:
296 iconv, zlib, libxml2, libxslt.
300 <section id="installation.windows.yaz">
303 <ulink url="&url.yaz;">YAZ</ulink> can be downloaded
305 <ulink url="&url.yaz.download.win32;">here</ulink>.
309 <section id="installation.windows.yazplusplus">
312 Get <ulink url="&url.yazplusplus;">YAZ++</ulink> as well.
313 Version 1.0 or later is required. For now get it from
315 <ulink url="&url.snapshot.download;">Snapshot area</ulink>.
318 YAZ++ includes NMAKE makefiles, similar to those found in the
323 <section id="installation.windows.metaproxy">
324 <title>Metaproxy</title>
326 Metaproxy is shipped with NMAKE makfiles as well - similar
327 to those found in the YAZ++/YAZ packages. Adjust this Makefile
328 to point to the proper locations of Boost, Libxslt, Libxml2,
329 zlib, iconv, yaz and yazpp.
332 After succesful compilation you'll find
333 <literal>metaproxy.exe</literal> in the
334 <literal>bin</literal> directory.
341 <chapter id="architecture">
342 <title>The Metaproxy Architecture</title>
344 The Metaproxy architecture is based on three concepts:
345 the <emphasis>package</emphasis>,
346 the <emphasis>route</emphasis>
347 and the <emphasis>filter</emphasis>.
351 <term>Packages</term>
354 A package is request or response, encoded in some protocol,
355 issued by a client, making its way through Metaproxy, send to or
356 received from a server, or sent back to the client.
359 The core of a package is the protocol unit - for example, a
360 Z39.50 Init Request or Search Response, or an SRU searchRetrieve
361 URL or Explain Response. In addition to this core, a package
362 also carries some extra information added and used by Metaproxy
366 In general, packages are doctored as they pass through
367 Metaproxy. For example, when the proxy performs authentication
368 and authorisation on a Z39.50 Init request, it removes the
369 authentication credentials from the package so that they are not
370 passed onto the back-end server; and when search-response
371 packages are obtained from multiple servers, they are merged
372 into a single unified package that makes its way back to the
381 Packages make their way through routes, which can be thought of
382 as programs that operate on the package data-type. Each
383 incoming package initially makes its way through a default
384 route, but may be switched to a different route based on various
385 considerations. Routes are made up of sequences of filters (see
394 Filters provide the individual instructions within a route, and
395 effect the necessary transformations on packages. A particular
396 configuration of Metaproxy is essentially a set of filters,
397 described by configuration details and arranged in order in one
398 or more routes. There are many kinds of filter - about a dozen
399 at the time of writing with more appearing all the time - each
400 performing a specific function and configured by different
404 The word ``filter'' is sometimes used rather loosely, in two
405 different ways: it may be used to mean a particular
406 <emphasis>type</emphasis> of filter, as when we speak of ``the
407 auth_simplefilter'' or ``the multi filter''; or it may be used
408 to be a specific <emphasis>instance</emphasis> of a filter
409 within a Metaproxy configuration. For example, a single
410 configuration will often contain multiple instances of the
411 <literal>z3950_client</literal> filter. In
412 operational terms, of these is a separate filter. In practice,
413 context always make it clear which sense of the word ``filter''
417 Extensibility of Metaproxy is primarily through the creation of
418 plugins that provide new filters. The filter API is small and
419 conceptually simple, but there are many details to master. See
421 <link linkend="extensions">extensions</link>.
427 Since packages are created and handled by the system itself, and
428 routes are conceptually simple, most of the remainder of this
429 document concentrates on filters. After a brief overview of the
430 filter types follows, along with some thoughts on possible future
437 <chapter id="filters">
438 <title>Filters</title>
442 <title>Introductory notes</title>
444 It's useful to think of Metaproxy as an interpreter providing a small
445 number of primitives and operations, but operating on a very
446 complex data type, namely the ``package''.
449 A package represents a Z39.50 or SRU/W request (whether for Init,
450 Search, Scan, etc.) together with information about where it came
451 from. Packages are created by front-end filters such as
452 <literal>frontend_net</literal> (see below), which reads them from
453 the network; other front-end filters are possible. They then pass
454 along a route consisting of a sequence of filters, each of which
455 transforms the package and may also have side-effects such as
456 generating logging. Eventually, the route will yield a response,
457 which is sent back to the origin.
460 There are many kinds of filter: some that are defined statically
461 as part of Metaproxy, and others may be provided by third parties
462 and dynamically loaded. They all conform to the same simple API
463 of essentially two methods: <function>configure()</function> is
464 called at startup time, and is passed a DOM tree representing that
465 part of the configuration file that pertains to this filter
466 instance: it is expected to walk that tree extracting relevant
467 information; and <function>process()</function> is called every
468 time the filter has to processes a package.
471 While all filters provide the same API, there are different modes
472 of functionality. Some filters are sources: they create
474 (<literal>frontend_net</literal>);
475 others are sinks: they consume packages and return a result
476 (<literal>z3950_client</literal>,
477 <literal>backend_test</literal>,
478 <literal>http_file</literal>);
479 the others are true filters, that read, process and pass on the
480 packages they are fed
481 (<literal>auth_simple</literal>,
482 <literal>log</literal>,
483 <literal>multi</literal>,
484 <literal>query_rewrite</literal>,
485 <literal>session_shared</literal>,
486 <literal>template</literal>,
487 <literal>virt_db</literal>).
492 <section id="overview.filter.types">
493 <title>Overview of filter types</title>
495 We now briefly consider each of the types of filter supported by
496 the core Metaproxy binary. This overview is intended to give a
497 flavour of the available functionality; more detailed information
498 about each type of filter is included below in
499 <link linkend="filterref"
500 >the reference guide to Metaproxy filters</link>.
503 The filters are here named by the string that is used as the
504 <literal>type</literal> attribute of a
505 <literal><filter></literal> element in the configuration
506 file to request them, with the name of the class that implements
507 them in parentheses. (The classname is not needed for normal
508 configuration and use of Metaproxy; it is useful only to
512 The filters are here listed in alphabetical order:
516 <title><literal>auth_simple</literal>
517 (mp::filter::AuthSimple)</title>
519 Simple authentication and authorisation. The configuration
520 specifies the name of a file that is the user register, which
521 lists <varname>username</varname>:<varname>password</varname>
522 pairs, one per line, colon separated. When a session begins, it
523 is rejected unless username and passsword are supplied, and match
524 a pair in the register. The configuration file may also specific
525 the name of another file that is the target register: this lists
526 lists <varname>username</varname>:<varname>dbname</varname>,<varname>dbname</varname>...
527 sets, one per line, with multiple database names separated by
528 commas. When a search is processed, it is rejected unless the
529 database to be searched is one of those listed as available to
535 <title><literal>backend_test</literal>
536 (mp::filter::Backend_test)</title>
538 A sink that provides dummy responses in the manner of the
539 <literal>yaz-ztest</literal> Z39.50 server. This is useful only
540 for testing. Seriously, you don't need this. Pretend you didn't
541 even read this section.
546 <title><literal>frontend_net</literal>
547 (mp::filter::FrontendNet)</title>
549 A source that accepts Z39.50 connections from a port
550 specified in the configuration, reads protocol units, and
551 feeds them into the next filter in the route. When the result is
552 revceived, it is returned to the original origin.
557 <title><literal>http_file</literal>
558 (mp::filter::HttpFile)</title>
560 A sink that returns the contents of files from the local
561 filesystem in response to HTTP requests. (Yes, Virginia, this
562 does mean that Metaproxy is also a Web-server in its spare time. So
563 far it does not contain either an email-reader or a Lisp
564 interpreter, but that day is surely coming.)
569 <title><literal>log</literal>
570 (mp::filter::Log)</title>
572 Writes logging information to standard output, and passes on
573 the package unchanged.
578 <title><literal>multi</literal>
579 (mp::filter::Multi)</title>
581 Performs multi-database searching.
583 <link linkend="multidb">the extended discussion</link>
584 of virtual databases and multi-database searching below.
589 <title><literal>query_rewrite</literal>
590 (mp::filter::QueryRewrite)</title>
592 Rewrites Z39.50 Type-1 and Type-101 (``RPN'') queries by a
593 three-step process: the query is transliterated from Z39.50
594 packet structures into an XML representation; that XML
595 representation is transformed by an XSLT stylesheet; and the
596 resulting XML is transliterated back into the Z39.50 packet
602 <title><literal>session_shared</literal>
603 (mp::filter::SessionShared)</title>
605 When this is finished, it will implement global sharing of
606 result sets (i.e. between threads and therefore between
607 clients), yielding performance improvements especially when
608 incoming requests are from a stateless environment such as a
609 web-server, in which the client process representing a session
610 might be any one of many. However:
614 This filter is not yet completed.
620 <title><literal>template</literal>
621 (mp::filter::Template)</title>
623 Does nothing at all, merely passing the packet on. (Maybe it
624 should be called <literal>nop</literal> or
625 <literal>passthrough</literal>?) This exists not to be used, but
626 to be copied - to become the skeleton of new filters as they are
627 written. As with <literal>backend_test</literal>, this is not
628 intended for civilians.
633 <title><literal>virt_db</literal>
634 (mp::filter::Virt_db)</title>
636 Performs virtual database selection: based on the name of the
637 database in the search request, a server is selected, and its
638 address added to the request in a <literal>VAL_PROXY</literal>
639 otherInfo packet. It will subsequently be used by a
640 <literal>z3950_client</literal> filter.
642 <link linkend="multidb">the extended discussion</link>
643 of virtual databases and multi-database searching below.
648 <title><literal>z3950_client</literal>
649 (mp::filter::Z3950Client)</title>
651 Performs Z39.50 searching and retrieval by proxying the
652 packages that are passed to it. Init requests are sent to the
653 address specified in the <literal>VAL_PROXY</literal> otherInfo
654 attached to the request: this may have been specified by client,
655 or generated by a <literal>virt_db</literal> filter earlier in
656 the route. Subsequent requests are sent to the same address,
657 which is remembered at Init time in a Session object.
663 <section id="future.directions">
664 <title>Future directions</title>
666 Some other filters that do not yet exist, but which would be
667 useful, are briefly described. These may be added in future
668 releases (or may be created by third parties, as loadable
674 <term><literal>frontend_cli</literal> (source)</term>
677 Command-line interface for generating requests.
682 <term><literal>frontend_sru</literal> (source)</term>
685 Receive SRU (and perhaps SRW) requests.
690 <term><literal>sru2z3950</literal> (filter)</term>
693 Translate SRU requests into Z39.50 requests.
698 <term><literal>sru_client</literal> (sink)</term>
701 SRU searching and retrieval.
706 <term><literal>srw_client</literal> (sink)</term>
709 SRW searching and retrieval.
714 <term><literal>opensearch_client</literal> (sink)</term>
717 A9 OpenSearch searching and retrieval.
727 <chapter id="configuration">
728 <title>Configuration: the Metaproxy configuration file format</title>
732 <title>Introductory notes</title>
734 If Metaproxy is an interpreter providing operations on packages, then
735 its configuration file can be thought of as a program for that
736 interpreter. Configuration is by means of a single file, the name
737 of which is supplied as the sole command-line argument to the
738 <command>metaproxy</command> program. (See
739 <link linkend="progref">the reference guide</link>
740 below for more information on invoking Metaproxy.)
743 The configuration files are written in XML. (But that's just an
744 implementation detail - they could just as well have been written
745 in YAML or Lisp-like S-expressions, or in a custom syntax.)
748 Since XML has been chosen, an XML schema,
749 <filename>config.xsd</filename>, is provided for validating
750 configuration files. This file is supplied in the
751 <filename>etc</filename> directory of the Metaproxy distribution. It
752 can be used by (among other tools) the <command>xmllint</command>
753 program supplied as part of the <literal>libxml2</literal>
757 xmllint --noout --schema etc/config.xsd my-config-file.xml
760 (A recent version of <literal>libxml2</literal> is required, as
761 support for XML Schemas is a relatively recent addition.)
765 <section id="overview.xml.structure">
766 <title>Overview of XML structure</title>
768 All elements and attributes are in the namespace
769 <ulink url="http://indexdata.dk/yp2/config/1"/>.
770 This is most easily achieved by setting the default namespace on
771 the top-level element, as here:
774 <yp2 xmlns="http://indexdata.dk/yp2/config/1">
777 The top-level element is <yp2>. This contains a
778 <start> element, a <filters> element and a
779 <routes> element, in that order. <filters> is
780 optional; the other two are mandatory. All three are
784 The <start> element is empty, but carries a
785 <literal>route</literal> attribute, whose value is the name of
786 route at which to start running - analogous to the name of the
787 start production in a formal grammar.
790 If present, <filters> contains zero or more <filter>
791 elements. Each filter carries a <literal>type</literal> attribute
792 which specifies what kind of filter is being defined
793 (<literal>frontend_net</literal>, <literal>log</literal>, etc.)
794 and contain various elements that provide suitable configuration
795 for a filter of its type. The filter-specific elements are
797 <link linkend="filterref">the reference guide below</link>.
798 Filters defined in this part of the file must carry an
799 <literal>id</literal> attribute so that they can be referenced
803 <routes> contains one or more <route> elements, each
804 of which must carry an <literal>id</literal> element. One of the
805 routes must have the ID value that was specified as the start
806 route in the <start> element's <literal>route</literal>
807 attribute. Each route contains zero or more <filter>
808 elements. These are of two types. They may be empty, but carry a
809 <literal>refid</literal> attribute whose value is the same as the
810 <literal>id</literal> of a filter previously defined in the
811 <filters> section. Alternatively, a route within a filter
812 may omit the <literal>refid</literal> attribute, but contain
813 configuration elements similar to those used for filters defined
814 in the <filters> section. (In other words, each filter in a
815 route may be included either by reference or by physical
821 <section id="example.configuration">
822 <title>An example configuration</title>
824 The following is a small, but complete, Metaproxy configuration
825 file (included in the distribution as
826 <literal>metaproxy/etc/config0.xml</literal>).
827 This file defines a very simple configuration that simply proxies
828 to whatever back-end server the client requests, but logs each
829 request and response. This can be useful for debugging complex
830 client-server dialogues.
832 <screen><![CDATA[<?xml version="1.0"?>
833 <yp2 xmlns="http://indexdata.dk/yp2/config/1">
834 <start route="start"/>
836 <filter id="frontend" type="frontend_net">
839 <filter id="backend" type="z3950_client">
844 <filter refid="frontend"/>
846 <filter refid="backend"/>
852 It works by defining a single route, called
853 <literal>start</literal>, which consists of a sequence of three
854 filters. The first and last of these are included by reference:
855 their <literal><filter></literal> elements have
856 <literal>refid</literal> attributes that refer to filters defined
857 within the prior <literal><filters></literal> section. The
858 middle filter is included inline in the route.
861 The three filters in the route are as follows: first, a
862 <literal>frontend_net</literal> filter accepts Z39.50 requests
863 from any host on port 9000; then these requests are passed through
864 a <literal>log</literal> filter that emits a message for each
865 request; they are then fed into a <literal>z3950_client</literal>
866 filter, which forwards the requests to the client-specified
867 back-end Z39.509 server. When the response arrives, it is handed
868 back to the <literal>log</literal> filter, which emits another
869 message; and then to the front-end filter, which returns the
870 response to the client.
877 <chapter id="multidb">
878 <title>Virtual databases and multi-database searching</title>
882 <title>Introductory notes</title>
884 Two of Metaproxy's filters are concerned with multiple-database
885 operations. Of these, <literal>virt_db</literal> can work alone
886 to control the routing of searches to one of a number of servers,
887 while <literal>multi</literal> can work together with
888 <literal>virt_db</literal> to perform multi-database searching, merging
889 the results into a unified result-set - ``metasearch in a box''.
892 The interaction between
893 these two filters is necessarily complex: it reflects the real,
894 irreducible complexity of multi-database searching in a protocol such
895 as Z39.50 that separates initialisation from searching, and in
896 which the database to be searched is not known at initialisation
900 It's possible to use these filters without understanding the
901 details of their functioning and the interaction between them; the
902 next two sections of this chapter are ``HOWTO'' guides for doing
903 just that. However, debugging complex configurations will require
904 a deeper understanding, which the last two sections of this
905 chapters attempt to provide.
910 <section id="multidb.virt_db">
911 <title>Virtual databases with the <literal>virt_db</literal> filter</title>
913 Working alone, the purpose of the
914 <literal>virt_db</literal>
915 filter is to route search requests to one of a selection of
916 back-end databases. In this way, a single Z39.50 endpoint
917 (running Metaproxy) can provide access to several different
918 underlying services, including those that would otherwise be
919 inaccessible due to firewalls. In many useful configurations, the
920 back-end databases are local to the Metaproxy installation, but
921 the software does not enforce this, and any valid Z39.50 servers
922 may be used as back-ends.
925 For example, a <literal>virt_db</literal>
926 filter could be set up so that searches in the virtual database
927 ``lc'' are forwarded to the Library of Congress bibliographic
928 catalogue server, and searches in the virtual database ``marc''
929 are forwarded to the toy database of MARC records that Index Data
930 hosts for testing purposes. A <literal>virt_db</literal>
931 configuration to make this switch would look like this:
933 <screen><![CDATA[<filter type="virt_db">
935 <database>lc</database>
936 <target>z3950.loc.gov:7090/voyager</target>
939 <database>marc</database>
940 <target>indexdata.dk/marc</target>
942 </filter>]]></screen>
944 As well as being useful in it own right, this filter also provides
945 the foundation for multi-database searching.
950 <section id="multidb.multi">
951 <title>Multi-database search with the <literal>multi</literal> filter</title>
953 To arrange for Metaproxy to broadcast searches to multiple back-end
954 servers, the configuration needs to include two components: a
955 <literal>virt_db</literal>
956 filter that specifies multiple
957 <literal><target></literal>
958 elements, and a subsequent
959 <literal>multi</literal>
960 filter. Here, for example, is a complete configuration that
961 broadcasts searches to both the Library of Congress catalogue and
962 Index Data's tiny testing database of MARC records:
964 <screen><![CDATA[<?xml version="1.0"?>
965 <yp2 xmlns="http://indexdata.dk/yp2/config/1">
966 <start route="start"/>
969 <filter type="frontend_net">
970 <threads>10</threads>
973 <filter type="virt_db">
975 <database>lc</database>
976 <target>z3950.loc.gov:7090/voyager</target>
979 <database>marc</database>
980 <target>indexdata.dk/marc</target>
983 <database>all</database>
984 <target>z3950.loc.gov:7090/voyager</target>
985 <target>indexdata.dk/marc</target>
988 <filter type="multi"/>
989 <filter type="z3950_client">
990 <timeout>30</timeout>
997 <literal>virt_db</literal>
998 filter that specifies multiple
999 <literal><target></literal>
1000 elements but without a subsequent
1001 <literal>multi</literal>
1002 filter yields surprising and undesirable results, as will be
1003 described below. Don't do that.)
1006 Metaproxy can be invoked with this configuration as follows:
1008 <screen>../src/metaproxy --config config-simple-multi.xml</screen>
1010 And thereafter, Z39.50 clients can connect to the running server
1011 (on port 9000, as specified in the configuration) and search in
1012 any of the databases
1013 <literal>lc</literal> (the Library of Congress catalogue),
1014 <literal>marc</literal> (Index Data's test database of MARC records)
1016 <literal>all</literal> (both of these). As an example, a session
1017 using the YAZ command-line client <literal>yaz-client</literal> is
1018 here included (edited for brevity and clarity):
1020 <screen><![CDATA[$ yaz-client @:9000
1024 Search was a success.
1025 Number of hits: 10000, setno 1
1029 Search was a success.
1030 Number of hits: 10, setno 3
1034 Search was a success.
1035 Number of hits: 10010, setno 4
1038 [marc]Record type: USmarc
1041 005 00000000000000.0
1042 008 910710c19910701nju 00010 eng
1046 100 10 $a Jack Collins
1047 245 10 $a How to program a computer
1053 [VOYAGER]Record type: USmarc
1055 005 20041229102447.0
1056 008 030910s2004 caua 000 0 eng
1057 035 $a (DLC) 2003112666
1058 906 $a 7 $b cbc $c orignew $d 4 $e epcn $f 20 $g y-gencatlg
1059 925 0 $a acquire $b 1 shelf copy $x policy default
1060 955 $a pc10 2003-09-10 $a pv12 2004-06-23 to SSCD; $h sj05 2004-11-30 $e sj05 2004-11-30 to Shelf.
1063 040 $a DLC $c DLC $d DLC
1064 050 00 $a MLCM 2004/03312 (G)
1065 245 10 $a 007, everything or nothing : $b Prima's official strategy guide / $c created by Kaizen Media Group.
1066 246 3 $a Double-O-seven, everything or nothing
1067 246 30 $a Prima's official strategy guide
1068 260 $a Roseville, CA : $b Prima Games, $c c2004.
1069 300 $a 161 p. : $b col. ill. ; $c 28 cm.
1070 500 $a "Platforms: Nintendo GameCube, Macintosh, PC, PlayStation 2 computer entertainment system, Xbox"--P. [4] of cover.
1071 650 0 $a Video games.
1072 710 2 $a Kaizen Media Group.
1073 856 42 $3 Publisher description $u http://www.loc.gov/catdir/description/random052/2003112666.html
1078 As can be seen, the first record in the result set is from the
1079 Index Data test database, and the second from the Library of
1080 Congress database. The result-set continues alternating records
1081 round-robin style until the point where one of the databases'
1082 records are exhausted.
1085 This example uses only two back-end databases; more may be used.
1086 There is no limitation imposed on the number of databases that may
1087 be metasearched in this way: issues of resource usage and
1088 administrative complexity dictate the practical limits.
1093 <section id="multidb.what">
1094 <title>What's going on?</title>
1096 <title>Lark's vomit</title>
1098 This section goes into a level of technical detail that is
1099 probably not necessary in order to configure and use Metaproxy.
1100 It is provided only for those who like to know how things work.
1101 You should feel free to skip on to the next section if this one
1102 doesn't seem like fun.
1106 Hold on tight - this may get a little hairy.
1109 In the general course of things, a Z39.50 Init request may carry
1110 with it an otherInfo packet of type <literal>VAL_PROXY</literal>,
1111 whose value indicates the address of a Z39.50 server to which the
1112 ultimate connection is to be made. (This otherInfo packet is
1113 supported by YAZ-based Z39.50 clients and servers, but has not yet
1114 been ratified by the Maintenance Agency and so is not widely used
1115 in non-Index Data software. We're working on it.)
1116 The <literal>VAL_PROXY</literal> packet functions
1117 analogously to the absoluteURI-style Request-URI used with the GET
1118 method when a web browser asks a proxy to forward its request: see
1120 <ulink url="http://www.w3.org/Protocols/rfc2616/rfc2616-sec5.html#sec5.1.2"
1121 >Request-URI</ulink>
1123 <ulink url="http://www.w3.org/Protocols/rfc2616/rfc2616.html"
1124 >the HTTP 1.1 specification</ulink>.
1127 The role of the <literal>virt_db</literal> filter is to rewrite
1128 this otherInfo packet dependent on the virtual database that the
1129 client wants to search.
1132 When Metaproxy receives a Z39.50 Init request from a client, it
1133 doesn't immediately forward that request to the back-end server.
1134 Why not? Because it doesn't know <emphasis>which</emphasis>
1135 back-end server to forward it to until the client sends a search
1136 request that specifies the database that it wants to search in.
1137 Instead, it just treasures the Init request up in its heart; and,
1138 later, the first time the client does a search on one of the
1139 specified virtual databases, a connection is forged to the
1140 appropriate server and the Init request is forwarded to it. If,
1141 later in the session, the same client searches in a different
1142 virtual database, then a connection is forged to the server that
1143 hosts it, and the same cached Init request is forwarded there,
1147 All of this clever Init-delaying is done by the
1148 <literal>frontend_net</literal> filter. The
1149 <literal>virt_db</literal> filter knows nothing about it; in
1150 fact, because the Init request that is received from the client
1151 doesn't get forwarded until a Search request is received, the
1152 <literal>virt_db</literal> filter (and the
1153 <literal>z3950_client</literal> filter behind it) doesn't even get
1154 invoked at Init time. The <emphasis>only</emphasis> thing that a
1155 <literal>virt_db</literal> filter ever does is rewrite the
1156 <literal>VAL_PROXY</literal> otherInfo in the requests that pass
1160 ### Describe the use of multiple <literal>VAL_PROXY</literal>
1161 otherInfos, added by <literal>virt_db</literal> and used by
1162 <literal>multi</literal>.
1167 <section id="multidb.picture">
1168 <title>A picture is worth a thousand words (but only five hundred on 64-bit architectures)</title>
1172 <imagedata fileref="multi.pdf" format="PDF"/>
1175 <imagedata fileref="multi.png" format="PNG"/>
1178 <!-- Fall back if none of the images can be used -->
1180 [Here there should be a diagram showing the progress of
1181 packages through the filters during a simple virtual-database
1182 search and a multi-database search, but is seems that your
1183 toolchain has not been able to include the diagram in this
1184 document. This is because of LaTeX suckage. Time to move to
1185 OpenOffice. Yes, really.]
1188 <!-- ### This used to work with an older version of DocBook
1190 <para>Caption: progress of packages through filters.</para>
1193 </inlinemediaobject>
1200 <chapter id="extensions">
1201 <title>Writing extensions for Metaproxy</title>
1202 <para>### To be written</para>
1208 <chapter id="classes">
1209 <title>Classes in the Metaproxy source code</title>
1213 <title>Introductory notes</title>
1215 <emphasis>Stop! Do not read this!</emphasis>
1216 You won't enjoy it at all. You should just skip ahead to
1217 <link linkend="refguide">the reference guide</link>,
1219 <!-- The remainder of this paragraph is lifted verbatim from
1220 Douglas Adams' _Hitch Hiker's Guide to the Galaxy_, chapter 8 -->
1221 you things you really need to know, like the fact that the
1222 fabulously beautiful planet Bethselamin is now so worried about
1223 the cumulative erosion by ten billion visiting tourists a year
1224 that any net imbalance between the amount you eat and the amount
1225 you excrete whilst on the planet is surgically removed from your
1226 bodyweight when you leave: so every time you go to the lavatory it
1227 is vitally important to get a receipt.
1230 This chapter contains documentation of the Metaproxy source code, and is
1231 of interest only to maintainers and developers. If you need to
1232 change Metaproxy's behaviour or write a new filter, then you will most
1233 likely find this chapter helpful. Otherwise it's a waste of your
1234 good time. Seriously: go and watch a film or something.
1235 <citetitle>This is Spinal Tap</citetitle> is particularly good.
1238 Still here? OK, let's continue.
1241 In general, classes seem to be named big-endianly, so that
1242 <literal>FactoryFilter</literal> is not a filter that filters
1243 factories, but a factory that produces filters; and
1244 <literal>FactoryStatic</literal> is a factory for the statically
1245 registered filters (as opposed to those that are dynamically
1250 <section id="individual.classes">
1251 <title>Individual classes</title>
1253 The classes making up the Metaproxy application are here listed by
1254 class-name, with the names of the source files that define them in
1259 <title><literal>mp::FactoryFilter</literal>
1260 (<filename>factory_filter.cpp</filename>)</title>
1262 A factory class that exists primarily to provide the
1263 <literal>create()</literal> method, which takes the name of a
1264 filter class as its argument and returns a new filter of that
1265 type. To enable this, the factory must first be populated by
1266 calling <literal>add_creator()</literal> for static filters (this
1267 is done by the <literal>FactoryStatic</literal> class, see below)
1268 and <literal>add_creator_dyn()</literal> for filters loaded
1274 <title><literal>mp::FactoryStatic</literal>
1275 (<filename>factory_static.cpp</filename>)</title>
1277 A subclass of <literal>FactoryFilter</literal> which is
1278 responsible for registering all the statically defined filter
1279 types. It does this by knowing about all those filters'
1280 structures, which are listed in its constructor. Merely
1281 instantiating this class registers all the static classes. It is
1282 for the benefit of this class that <literal>struct
1283 metaproxy_1_filter_struct</literal> exists, and that all the filter
1284 classes provide a static object of that type.
1289 <title><literal>mp::filter::Base</literal>
1290 (<filename>filter.cpp</filename>)</title>
1292 The virtual base class of all filters. The filter API is, on the
1293 surface at least, extremely simple: two methods.
1294 <literal>configure()</literal> is passed a DOM tree representing
1295 that part of the configuration file that pertains to this filter
1296 instance, and is expected to walk that tree extracting relevant
1297 information. And <literal>process()</literal> processes a
1298 package (see below). That surface simplicitly is a bit
1299 misleading, as <literal>process()</literal> needs to know a lot
1300 about the <literal>Package</literal> class in order to do
1306 <title><literal>mp::filter::AuthSimple</literal>,
1307 <literal>Backend_test</literal>, etc.
1308 (<filename>filter_auth_simple.cpp</filename>,
1309 <filename>filter_backend_test.cpp</filename>, etc.)</title>
1311 Individual filters. Each of these is implemented by a header and
1312 a source file, named <filename>filter_*.hpp</filename> and
1313 <filename>filter_*.cpp</filename> respectively. All the header
1314 files should be pretty much identical, in that they declare the
1315 class, including a private <literal>Rep</literal> class and a
1316 member pointer to it, and the two public methods. The only extra
1317 information in any filter header is additional private types and
1318 members (which should really all be in the <literal>Rep</literal>
1319 anyway) and private methods (which should also remain known only
1320 to the source file, but C++'s brain-damaged design requires this
1321 dirty laundry to be exhibited in public. Thanks, Bjarne!)
1324 The source file for each filter needs to supply:
1329 A definition of the private <literal>Rep</literal> class.
1334 Some boilerplate constructors and destructors.
1339 A <literal>configure()</literal> method that uses the
1340 appropriate XML fragment.
1345 Most important, the <literal>process()</literal> method that
1346 does all the actual work.
1353 <title><literal>mp::Package</literal>
1354 (<filename>package.cpp</filename>)</title>
1356 Represents a package on its way through the series of filters
1357 that make up a route. This is essentially a Z39.50 or SRU APDU
1358 together with information about where it came from, which is
1359 modified as it passes through the various filters.
1364 <title><literal>mp::Pipe</literal>
1365 (<filename>pipe.cpp</filename>)</title>
1367 This class provides a compatibility layer so that we have an IPC
1368 mechanism that works the same under Unix and Windows. It's not
1369 particularly exciting.
1374 <title><literal>mp::RouterChain</literal>
1375 (<filename>router_chain.cpp</filename>)</title>
1382 <title><literal>mp::RouterFleXML</literal>
1383 (<filename>router_flexml.cpp</filename>)</title>
1390 <title><literal>mp::Session</literal>
1391 (<filename>session.cpp</filename>)</title>
1398 <title><literal>mp::ThreadPoolSocketObserver</literal>
1399 (<filename>thread_pool_observer.cpp</filename>)</title>
1406 <title><literal>mp::util</literal>
1407 (<filename>util.cpp</filename>)</title>
1409 A namespace of various small utility functions and classes,
1410 collected together for convenience. Most importantly, includes
1411 the <literal>mp::util::odr</literal> class, a wrapper for YAZ's
1417 <title><literal>mp::xml</literal>
1418 (<filename>xmlutil.cpp</filename>)</title>
1420 A namespace of various XML utility functions and classes,
1421 collected together for convenience.
1427 <section id="other.source.files">
1428 <title>Other Source Files</title>
1430 In addition to the Metaproxy source files that define the classes
1431 described above, there are a few additional files which are
1432 briefly described here:
1436 <term><literal>metaproxy_prog.cpp</literal></term>
1439 The main function of the <command>metaproxy</command> program.
1444 <term><literal>ex_router_flexml.cpp</literal></term>
1447 Identical to <literal>metaproxy_prog.cpp</literal>: it's not clear why.
1452 <term><literal>test_*.cpp</literal></term>
1455 Unit-tests for various modules.
1461 ### Still to be described:
1462 <literal>ex_filter_frontend_net.cpp</literal>,
1463 <literal>filter_dl.cpp</literal>,
1464 <literal>plainfile.cpp</literal>,
1465 <literal>tstdl.cpp</literal>.
1472 <chapter id="refguide">
1473 <title>Reference guide</title>
1475 The material in this chapter is drawn directly from the individual
1476 manual entries. In particular, the Metaproxy invocation section is
1477 available using <command>man metaproxy</command>, and the section
1478 on each individual filter is available using the name of the filter
1479 as the argument to the <command>man</command> command.
1483 <section id="progref">
1484 <title>Metaproxy invocation</title>
1489 <section id="filterref">
1490 <title>Reference guide to Metaproxy filters</title>
1497 <!-- Keep this comment at the end of the file
1502 sgml-minimize-attributes:nil
1503 sgml-always-quote-attributes:t
1506 sgml-parent-document: "main.xml"
1507 sgml-local-catalogs: nil
1508 sgml-namecase-general:t