1 <?xml version="1.0" standalone="no"?>
2 <!DOCTYPE book PUBLIC "-//OASIS//DTD DocBook XML V4.1//EN"
3 "http://www.oasis-open.org/docbook/xml/4.1/docbookx.dtd"
5 <!ENTITY % local SYSTEM "local.ent">
7 <!ENTITY manref SYSTEM "manref.xml">
8 <!ENTITY progref SYSTEM "progref.xml">
9 <!ENTITY % common SYSTEM "common/common.ent">
11 <!-- Next line allows imagedata/@format="PDF" and is taken from
12 http://lists.oasis-open.org/archives/docbook/200303/msg00163.html
14 <!ENTITY % local.notation.class "| PDF">
15 <!-- Next line is necessary for some XML parsers, for reasons I
16 don't understand. I got this from
17 http://lists.oasis-open.org/archives/docbook/200303/msg00180.html
19 <!NOTATION PDF SYSTEM "PDF">
21 <!-- $Id: book.xml,v 1.56 2007-02-17 12:28:11 adam Exp $ -->
24 <title>Metaproxy - User's Guide and Reference</title>
27 <firstname>Adam</firstname><surname>Dickmeiss</surname>
30 <firstname>Marc</firstname><surname>Cromme</surname>
33 <firstname>Mike</firstname><surname>Taylor</surname>
36 <releaseinfo>&version;</releaseinfo>
38 <year>2005-2007</year>
39 <holder>Index Data ApS</holder>
43 This manual is part of Metaproxy version &version;.
46 Metaproxy is a universal router, proxy and encapsulated
47 metasearcher for information retrieval protocols. It accepts,
48 processes, interprets and redirects requests from IR clients using
49 standard protocols such as the binary
50 <ulink url="&url.z39.50;">ANSI/NISO Z39.50</ulink>
51 and the information search and retireval
52 web services <ulink url="&url.sru;">SRU</ulink>
53 and <ulink url="&url.srw;">SRW</ulink>, as
54 well as functioning as a limited
55 <ulink url="&url.http;">HTTP</ulink> server.
58 Metaproxy is configured by an XML file which
59 specifies how the software should function in terms of routes that
60 the request packets can take through the proxy, each step on a
61 route being an instantiation of a filter. Filters come in many
62 types, one for each operation: accepting Z39.50 packets, logging,
63 query transformation, multiplexing, etc. Further filter-types can
64 be added as loadable modules to extend Metaproxy functionality,
68 Metaproxy is <emphasis>not</emphasis> open-source software, but
69 may be freely downloaded, unpacked, inspected, built and run for
70 evaluation purposes. Deployment requires a separate, commercial,
76 <imagedata fileref="common/id.png" format="PNG"/>
79 <imagedata fileref="common/id.eps" format="EPS"/>
86 <chapter id="introduction">
87 <title>Introduction</title>
91 <ulink url="&url.metaproxy;">Metaproxy</ulink>
92 is a stand alone program that acts as a universal router, proxy and
93 encapsulated metasearcher for information retrieval protocols such
94 as <ulink url="&url.z39.50;">Z39.50</ulink>, and in the future
95 <ulink url="&url.sru;">SRU</ulink> and <ulink url="&url.srw;">SRW</ulink>.
96 To clients, it acts as a server of these protocols: it can be searched,
97 records can be retrieved from it, etc.
98 To servers, it acts as a client: it searches in them,
99 retrieves records from them, etc. it satisfies its clients'
100 requests by transforming them, multiplexing them, forwarding them
101 on to zero or more servers, merging the results, transforming
102 them, and delivering them back to the client. In addition, it
103 acts as a simple <ulink url="&url.http;">HTTP</ulink> server; support
104 for further protocols can be added in a modular fashion, through the
105 creation of new filters.
110 Fish, bananas, cold pyjamas,
111 Mutton, beef and trout!
112 - attributed to Cole Porter.
115 Metaproxy is a more capable alternative to
116 <ulink url="&url.yazproxy;">YAZ Proxy</ulink>,
117 being more powerful, flexible, configurable and extensible. Among
118 its many advantages over the older, more pedestrian work are
119 support for multiplexing (encapsulated metasearching), routing by
120 database name, authentication and authorization and serving local
121 files via HTTP. Equally significant, its modular architecture
122 facilitites the creation of pluggable modules implementing further
126 This manual will describe how to install Metaproxy
127 before giving an overview of its architecture, then discussing the
128 key concept of a filter in some depth and giving an overview of
129 the various filter types, then discussing the configuration file
130 format. After this come several optional chapters which may be
131 freely skipped: a detailed discussion of virtual databases and
132 multi-database searching, some notes on writing extensions
133 (additional filter types) and a high-level description of the
134 source code. Finally comes the reference guide, which contains
135 instructions for invoking the <command>metaproxy</command>
136 program, and detailed information on each type of filter,
141 <chapter id="license">
142 <title>The Metaproxy License</title>
143 <orderedlist numeration="arabic">
146 You are allowed to download this software for evaluation purposes.
147 You can unpack it, build it, run it, see how it works and how it fits
148 your needs, all at zero cost.
153 You may NOT deploy the software. For the purposes of this license,
154 deployment means running it for any purpose other than evaluation,
155 whether or not you or anyone else makes a profit from doing so. If
156 you wish to deploy the software, you must first contact Index Data and
157 arrange to purchase a DEPLOYMENT LICENCE. If you are unsure
158 whether or not your proposed use of the software constitutes
159 deployment, email us at <literal>info@indexdata.com</literal>
165 You may modify your copy of the software (fix bugs, add features)
166 if you need to. We encourage you to send your changes back to us for
167 integration into the master copy, but you are not obliged to do so. You
168 may NOT pass your changes on to any other party.
173 There is NO WARRANTY for this software, to the extent permitted by
174 applicable law. We provide the software ``as is'' without warranty of
175 any kind, either expressed or implied, including, but not limited to, the
176 implied warranties of MERCHANTABILITY and FITNESS FOR A
177 PARTICULAR PURPOSE. The entire risk as to the quality and
178 performance of the software is with you. Should the software prove
179 defective, you assume the cost of all necessary servicing, repair or
180 correction. In no event unless required by applicable law will we be
181 liable to you for damages, arising out of the use of the software,
182 including but not limited to loss of data or data being rendered
188 All rights to the software are reserved by Index Data except where
189 this license explicitly says otherwise.
195 <chapter id="installation">
196 <title>Installation</title>
198 Metaproxy depends on the following tools/libraries:
200 <varlistentry><term><ulink url="&url.yazplusplus;">YAZ++</ulink></term>
203 This is a C++ library based on <ulink url="&url.yaz;">YAZ</ulink>.
207 <varlistentry><term><ulink url="&url.libxslt;">Libxslt</ulink></term>
209 <para>This is an XSLT processor - based on
210 <ulink url="&url.libxml2;">Libxml2</ulink>. Both Libxml2 and
211 Libxslt must be installed with the development components
212 (header files, etc.) as well as the run-time libraries.
216 <varlistentry><term><ulink url="&url.boost;">Boost</ulink></term>
219 The popular C++ library. Initial versions of Metaproxy
220 was built with 1.33.0. Version 1.33.1 works too.
227 In order to compile Metaproxy a modern C++ compiler is
228 required. Boost, in particular, requires the C++ compiler
229 to facilitate the newest features. Refer to Boost
230 <ulink url="&url.boost.compilers.status;">Compiler Status</ulink>
231 for more information.
234 We have successfully built Metaproxy using the compilers
235 <ulink url="&url.gcc;">GCC</ulink> version 4.0 and
236 <ulink url="&url.vstudio;">Microsoft Visual Studio</ulink> 2003/2005.
239 <section id="installation.unix">
240 <title>Installation on Unix (from Source)</title>
242 Here is a quick step-by-step guide on how to compile all the
243 tools that Metaproxy uses. Only few systems have none of the required
244 tools binary packages. If, for example, Libxml2/libxslt are already
245 installed as development packages use those (and omit compilation).
252 gunzip -c libxml2-version.tar.gz|tar xf -
260 gunzip -c libxslt-version.tar.gz|tar xf -
271 gunzip -c yaz-version.tar.gz|tar xf -
279 gunzip -c yazpp-version.tar.gz|tar xf -
290 gunzip -c boost-version.tar.gz|tar xf -
301 gunzip -c metaproxy-version.tar.gz|tar xf -
310 <section id="installation.debian">
311 <title>Installation on Debian GNU/Linux</title>
313 All dependencies for Metaproxy are available as
314 <ulink url="&url.debian;">Debian</ulink>
315 packages for the sarge (stable in 2005) and etch (testing in 2005)
319 The procedures for Debian based systems, such as
320 <ulink url="&url.ubuntu;">Ubuntu</ulink> is probably similar
323 There is currently no official Debian package for YAZ++.
324 And the Debian package for YAZ is probably too old.
325 Update the <filename>/etc/apt/sources.list</filename>
326 to include the Index Data repository.
327 See YAZ' <ulink url="&url.yaz.download.debian;">Download Debian</ulink>
328 for more information.
331 apt-get install libxslt1-dev
332 apt-get install libyazpp-dev
333 apt-get install libboost-dev
334 apt-get install libboost-thread-dev
335 apt-get install libboost-date-time-dev
336 apt-get install libboost-program-options-dev
337 apt-get install libboost-test-dev
340 With these packages installed, the usual configure + make
341 procedure can be used for Metaproxy as outlined in
342 <xref linkend="installation.unix"/>.
346 <section id="installation.windows">
347 <title>Installation on Windows</title>
349 Metaproxy can be compiled with Microsoft
350 <ulink url="&url.vstudio;">Visual Studio</ulink>.
351 Version 2003 (C 7.1) and 2005 (C 8.0) is known to work.
353 <section id="installation.windows.boost">
356 Get Boost from its <ulink url="&url.boost;">home page</ulink>.
357 You also need Boost Jam (an alternative to make).
358 That's also available from the Boost home page.
359 The files to be downloaded are called something like:
360 <filename>boost_1_33-1.exe</filename>
362 <filename>boost-jam-3.1.12-1-ntx86.zip</filename>.
363 Unpack Boost Jam first. Put <filename>bjam.exe</filename>
364 in your system path. Make a command prompt and ensure
365 it can be found automatically. If not check the PATH.
366 The Boost .exe is a self-extracting exe with
367 complete source for Boost. Compile that source with
368 Boost Jam (An alternative to Make).
369 The compilation takes a while.
370 For Visual Studio 2003, use
372 bjam "-sTOOLS=vc-7_1"
374 Here <literal>vc-7_1</literal> refers to a "Toolset" (compiler system).
375 For Visual Studio 2005, use
377 bjam "-sTOOLS=vc-8_0"
379 To install the libraries in a common place, use
381 bjam "-sTOOLS=vc-7_1" install
383 (or vc-8_0 for VS 2005).
386 By default, the Boost build process installs the resulting
387 libraries + header files in
388 <literal>\boost\lib</literal>, <literal>\boost\include</literal>.
391 For more information about installing Boost refer to the
392 <ulink url="&url.boost.getting.started;">getting started</ulink>
397 <section id="installation.windows.libxslt">
398 <title>Libxslt</title>
400 <ulink url="&url.libxslt;">Libxslt</ulink> can be downloaded
402 <ulink url="&url.libxml2.download.win32;">here</ulink>.
405 Libxslt has other dependencies, but these can all be downloaded
406 from the same site. Get the following:
407 iconv, zlib, libxml2, libxslt.
411 <section id="installation.windows.yaz">
414 <ulink url="&url.yaz;">YAZ</ulink> can be downloaded
416 <ulink url="&url.yaz.download.win32;">here</ulink>.
420 <section id="installation.windows.yazplusplus">
423 Get <ulink url="&url.yazplusplus;">YAZ++</ulink> as well.
424 Version 1.0 or later is required. For now get it from
426 <ulink url="&url.snapshot.download;">Snapshot area</ulink>.
429 YAZ++ includes NMAKE makefiles, similar to those found in the
434 <section id="installation.windows.metaproxy">
435 <title>Metaproxy</title>
437 Metaproxy is shipped with NMAKE makefiles as well - similar
438 to those found in the YAZ++/YAZ packages. Adjust this Makefile
439 to point to the proper locations of Boost, Libxslt, Libxml2,
440 zlib, iconv, yaz and yazpp.
444 <varlistentry><term><literal>DEBUG</literal></term>
446 If set to 1, the software is
447 compiled with debugging libraries (code generation is
448 multi-threaded debug DLL).
449 If set to 0, the software is compiled with release libraries
450 (code generation is multi-threaded DLL).
455 <term><literal>BOOST</literal></term>
458 Boost install location
464 <term><literal>BOOST_VERSION</literal></term>
467 Boost version (replace . with _).
473 <term><literal>BOOST_TOOLSET</literal></term>
482 <term><literal>LIBXSLT_DIR</literal>,
483 <literal>LIBXML2_DIR</literal> ..</term>
486 Specify the locations of Libxslt, libiconv, libxml2 and
495 After successful compilation you'll find
496 <literal>metaproxy.exe</literal> in the
497 <literal>bin</literal> directory.
505 <chapter id="yazproxy-comparison">
506 <title>YAZ Proxy Comparison</title>
508 The table below lists facilities either supported by either
509 <ulink url="&url.yazproxy;">YAZ Proxy</ulink> or Metaproxy.
511 <table id="yazproxy-comparison-table">
512 <title>Metaproxy / YAZ Proxy comparison</title>
516 <entry>Facility</entry>
517 <entry>Metaproxy</entry>
518 <entry>YAZ Proxy</entry>
523 <entry>Z39.50 server</entry>
524 <entry>Using filter <literal>frontend_net</literal></entry>
525 <entry>Supported</entry>
528 <entry>SRU server</entry>
529 <entry>Supported with filter <literal>sru_z3950</literal></entry>
530 <entry>Supported</entry>
533 <entry>Z39.50 client</entry>
534 <entry>Supported with filter <literal>z3950_client</literal></entry>
535 <entry>Supported</entry>
538 <entry>SRU client</entry>
539 <entry>Unsupported</entry>
540 <entry>Unsupported</entry>
543 <entry>Connection reuse</entry>
544 <entry>Supported with filter <literal>session_shared</literal></entry>
545 <entry>Supported</entry>
548 <entry>Connection share</entry>
549 <entry>Supported with filter <literal>session_shared</literal></entry>
550 <entry>Unsupported</entry>
553 <entry>Result set reuse</entry>
554 <entry>Supported with filter <literal>session_shared</literal></entry>
555 <entry>Within one Z39.50 session / HTTP keep-alive</entry>
558 <entry>Record cache</entry>
559 <entry>Unsupported</entry>
560 <entry>Supported for last result set within one Z39.50/HTTP-keep alive session</entry>
563 <entry>Z39.50 Virtual database, i.e. select any Z39.50 target for database</entry>
564 <entry>Supported with filter <literal>virt_db</literal></entry>
565 <entry>Unsupported</entry>
568 <entry>SRU Virtual database, i.e. select any Z39.50 target for path</entry>
569 <entry>Supported with filter <literal>virt_db</literal>,
570 <literal>sru_z3950</literal></entry>
571 <entry>Supported</entry>
574 <entry>Multi target search</entry>
575 <entry>Supported with filter <literal>multi</literal> (round-robin)</entry>
576 <entry>Unsupported</entry>
579 <entry>Retrieval and search limits</entry>
580 <entry>Unsupported</entry>
581 <entry>Supported</entry>
584 <entry>Bandwidth limits</entry>
585 <entry>Unsupported</entry>
586 <entry>Supported</entry>
589 <entry>Connect limits</entry>
590 <entry>Unsupported</entry>
591 <entry>Supported</entry>
594 <entry>Retrieval sanity check and conversions</entry>
595 <entry>Supported using filter <literal>record_transform</literal></entry>
596 <entry>Supported</entry>
599 <entry>Query check</entry>
601 Supported in a limited way using <literal>query_rewrite</literal>
603 <entry>Supported</entry>
606 <entry>Query rewrite</entry>
607 <entry>Supported with <literal>query_rewrite</literal></entry>
608 <entry>Unsupported</entry>
611 <entry>Session invalidate for -1 hits</entry>
612 <entry>Unsupported</entry>
613 <entry>Supported</entry>
616 <entry>Architecture</entry>
617 <entry>Multi-threaded + select for networked modules such as
618 <literal>frontend_net</literal>)</entry>
619 <entry>Single-threaded using select</entry>
623 <entry>Extensability</entry>
624 <entry>Most functionality implemented as loadable modules</entry>
625 <entry>Unsupported and experimental</entry>
629 <entry><ulink url="&url.usemarcon;">USEMARCON</ulink></entry>
630 <entry>Unsupported</entry>
631 <entry>Supported</entry>
635 <entry>Portability</entry>
637 Requires YAZ, YAZ++ and modern C++ compiler supporting
638 <ulink url="&url.boost;">Boost</ulink>.
641 Requires YAZ and YAZ++.
642 STL is not required so pretty much any C++ compiler out there should work.
651 <chapter id="architecture">
652 <title>The Metaproxy Architecture</title>
654 The Metaproxy architecture is based on three concepts:
655 the <emphasis>package</emphasis>,
656 the <emphasis>route</emphasis>
657 and the <emphasis>filter</emphasis>.
661 <term>Packages</term>
664 A package is request or response, encoded in some protocol,
665 issued by a client, making its way through Metaproxy, send to or
666 received from a server, or sent back to the client.
669 The core of a package is the protocol unit - for example, a
670 Z39.50 Init Request or Search Response, or an SRU searchRetrieve
671 URL or Explain Response. In addition to this core, a package
672 also carries some extra information added and used by Metaproxy
676 In general, packages are doctored as they pass through
677 Metaproxy. For example, when the proxy performs authentication
678 and authorization on a Z39.50 Init request, it removes the
679 authentication credentials from the package so that they are not
680 passed onto the back-end server; and when search-response
681 packages are obtained from multiple servers, they are merged
682 into a single unified package that makes its way back to the
691 Packages make their way through routes, which can be thought of
692 as programs that operate on the package data-type. Each
693 incoming package initially makes its way through a default
694 route, but may be switched to a different route based on various
695 considerations. Routes are made up of sequences of filters (see
704 Filters provide the individual instructions within a route, and
705 effect the necessary transformations on packages. A particular
706 configuration of Metaproxy is essentially a set of filters,
707 described by configuration details and arranged in order in one
708 or more routes. There are many kinds of filter - about a dozen
709 at the time of writing with more appearing all the time - each
710 performing a specific function and configured by different
714 The word ``filter'' is sometimes used rather loosely, in two
715 different ways: it may be used to mean a particular
716 <emphasis>type</emphasis> of filter, as when we speak of ``the
717 auth_simple filter'' or ``the multi filter''; or it may be used
718 to be a specific <emphasis>instance</emphasis> of a filter
719 within a Metaproxy configuration. For example, a single
720 configuration will often contain multiple instances of the
721 <literal>z3950_client</literal> filter. In
722 operational terms, of these is a separate filter. In practice,
723 context always make it clear which sense of the word ``filter''
727 Extensibility of Metaproxy is primarily through the creation of
728 plugins that provide new filters. The filter API is small and
729 conceptually simple, but there are many details to master. See
731 <link linkend="filters">Filters</link>.
737 Since packages are created and handled by the system itself, and
738 routes are conceptually simple, most of the remainder of this
739 document concentrates on filters. After a brief overview of the
740 filter types follows, along with some thoughts on possible future
747 <chapter id="filters">
748 <title>Filters</title>
752 <title>Introductory notes</title>
754 It's useful to think of Metaproxy as an interpreter providing a small
755 number of primitives and operations, but operating on a very
756 complex data type, namely the ``package''.
759 A package represents a Z39.50 or SRU/W request (whether for Init,
760 Search, Scan, etc.) together with information about where it came
761 from. Packages are created by front-end filters such as
762 <literal>frontend_net</literal> (see below), which reads them from
763 the network; other front-end filters are possible. They then pass
764 along a route consisting of a sequence of filters, each of which
765 transforms the package and may also have side-effects such as
766 generating logging. Eventually, the route will yield a response,
767 which is sent back to the origin.
770 There are many kinds of filter: some that are defined statically
771 as part of Metaproxy, and others may be provided by third parties
772 and dynamically loaded. They all conform to the same simple API
773 of essentially two methods: <function>configure()</function> is
774 called at startup time, and is passed an XML DOM tree representing that
775 part of the configuration file that pertains to this filter
776 instance: it is expected to walk that tree extracting relevant
777 information; and <function>process()</function> is called every
778 time the filter has to processes a package.
781 While all filters provide the same API, there are different modes
782 of functionality. Some filters are sources: they create
784 (<literal>frontend_net</literal>);
785 others are sinks: they consume packages and return a result
786 (<literal>backend_test</literal>,
787 <literal>bounce</literal>,
788 <literal>http_file</literal>,
789 <literal>z3950_client</literal>);
790 the others are true filters, that read, process and pass on the
791 packages they are fed
792 (<literal>auth_simple</literal>,
793 <literal>log</literal>,
794 <literal>multi</literal>,
795 <literal>query_rewrite</literal>,
796 <literal>record_transform</literal>,
797 <literal>session_shared</literal>,
798 <literal>sru_z3950</literal>,
799 <literal>template</literal>,
800 <literal>virt_db</literal>).
805 <section id="overview.filter.types">
806 <title>Overview of filter types</title>
808 We now briefly consider each of the types of filter supported by
809 the core Metaproxy binary. This overview is intended to give a
810 flavor of the available functionality; more detailed information
811 about each type of filter is included below in
812 <xref linkend="reference"/>.
815 The filters are here named by the string that is used as the
816 <literal>type</literal> attribute of a
817 <literal><filter></literal> element in the configuration
818 file to request them, with the name of the class that implements
819 them in parentheses. (The classname is not needed for normal
820 configuration and use of Metaproxy; it is useful only to
824 The filters are here listed in alphabetical order:
831 New virt_db-alike that does inteligent peer choice, explain merging,
832 adds FD&N to explain. Keeps init responses (like "virt_db Classic"),
833 makes routing choices based on local explain knowledge. Ref IDDI
836 Filter to convert Explain Classic to ZeeRex.
838 CQL2PQF (which needs augmented ZeeRex) - MARC for Talis.
842 Figure out what additional information we need in:
843 ZeeRex (check against D3.1)
844 Init request (e.g. loop detection)
845 Query package (e.g. number of hops)
846 Query response (e.g. record source)
851 <title><literal>auth_simple</literal>
852 (mp::filter::AuthSimple)</title>
854 Simple authentication and authorization. The configuration
855 specifies the name of a file that is the user register, which
856 lists <varname>username</varname>:<varname>password</varname>
857 pairs, one per line, colon separated. When a session begins, it
858 is rejected unless username and passsword are supplied, and match
859 a pair in the register. The configuration file may also specific
860 the name of another file that is the target register: this lists
861 lists <varname>username</varname>:<varname>dbname</varname>,<varname>dbname</varname>...
862 sets, one per line, with multiple database names separated by
863 commas. When a search is processed, it is rejected unless the
864 database to be searched is one of those listed as available to
870 <title><literal>backend_test</literal>
871 (mp::filter::Backend_test)</title>
873 A partial sink that provides dummy responses in the manner of the
874 <literal>yaz-ztest</literal> Z39.50 server. This is useful only
875 for testing. Seriously, you don't need this. Pretend you didn't
876 even read this section.
881 <title><literal>bounce</literal>
882 (mp::filter::Bounce)</title>
884 A sink that swallows <emphasis>all packages</emphasis>,
885 and returns them almost unprocessed.
886 It never sends any package of any type further down the row, but
887 sets Z39.50 packages to Z_Close, and HTTP_Request packages to
888 HTTP_Response err code 400 packages, and adds a suitable bounce
890 The bounce filter is usually added at end of each filter chain route
891 to prevent infinite hanging of for example HTTP
892 requests packages when only the Z39.50 client partial sink
893 filter is found in the
899 <title><literal>cql_rpn</literal>
900 (mp::filter::CQLtoRPN)</title>
902 A query language transforming filter which catches Z39.50
903 <literal>searchRequest</literal>
904 packages containing <literal>CQL</literal> queries, transforms
905 those to <literal>RPN</literal> queries,
906 and sends the <literal>searchRequests</literal> on to the next
907 filters. It is among other things useful in a SRU context.
912 <title><literal>frontend_net</literal>
913 (mp::filter::FrontendNet)</title>
915 A source that accepts Z39.50 connections from a port
916 specified in the configuration, reads protocol units, and
917 feeds them into the next filter in the route. When the result is
918 received, it is returned to the original origin.
923 <title><literal>http_file</literal>
924 (mp::filter::HttpFile)</title>
926 A partial sink which swallows only
927 <literal>HTTP_Request</literal> packages, and
928 returns the contents of files from the local
929 filesystem in response to HTTP requests.
930 It lets Z39.50 packages and all other forthcoming package types
933 does mean that Metaproxy is also a Web-server in its spare time. So
934 far it does not contain either an email-reader or a Lisp
935 interpreter, but that day is surely coming.)
940 <title><literal>load_balance</literal>
941 (mp::filter::LoadBalance)</title>
943 Performs load balancing for incoming Z39.50 init requests.
944 It is used together with the <literal>virt_db</literal> filter,
945 but unlike the <literal>multi</literal> filter it does send an
946 entire session to only one of the virtual backends. The
947 <literal>load_balance</literal> filter is assuming that
948 all backend targets have equal content, and chooses the backend
949 with least load cost for a new session.
952 This filter is experimental and yet not mature for heavy load
960 <title><literal>log</literal>
961 (mp::filter::Log)</title>
963 Writes logging information to standard output, and passes on
964 the package unchanged. A log file name can be specified, as well
965 as multiple different logging formats.
970 <title><literal>multi</literal>
971 (mp::filter::Multi)</title>
973 Performs multi-database searching.
975 <link linkend="multidb">the extended discussion</link>
976 of virtual databases and multi-database searching below.
981 <title><literal>query_rewrite</literal>
982 (mp::filter::QueryRewrite)</title>
984 Rewrites Z39.50 <literal>Type-1</literal>
985 and <literal>Type-101</literal> (``<literal>RPN</literal>'')
987 three-step process: the query is transliterated from Z39.50
988 packet structures into an XML representation; that XML
989 representation is transformed by an XSLT stylesheet; and the
990 resulting XML is transliterated back into the Z39.50 packet
997 <title><literal>record_transform</literal>
998 (mp::filter::RecordTransform)</title>
1000 This filter acts only on Z3950 present requests, and let all
1001 other types of packages and requests pass untouched. It's use is
1002 twofold: blocking Z3950 present requests, which the backend
1003 server does not understand and can not honor, and transforming
1004 the present syntax and elementset name according to the rules
1005 specified, to fetch only existing record formats, and transform
1006 them on the fly to requested record syntaxes.
1011 <title><literal>session_shared</literal>
1012 (mp::filter::SessionShared)</title>
1014 This filter implements global sharing of
1015 result sets (i.e. between threads and therefore between
1016 clients), yielding performance improvements by clever resource
1022 <title><literal>sru_z3950</literal>
1023 (mp::filter::SRUtoZ3950)</title>
1025 This filter transforms valid
1026 SRU GET/POST/SOAP searchRetrieve requests to Z3950 init, search,
1027 and present requests, and wraps the
1028 received hit counts and XML records into suitable SRU response
1030 The <literal>sru_z3950</literal> filter processes also SRU
1031 GET/POST/SOAP explain requests, returning
1032 either the absolute minimum required by the standard, or a full
1033 pre-defined ZeeReX explain record.
1035 <ulink url="&url.zeerex.explain;">ZeeReX Explain</ulink>
1036 standard pages and the
1037 <ulink url="&url.sru.explain;">SRU Explain</ulink> pages
1038 for more information on the correct explain syntax.
1039 SRU scan requests are not supported yet.
1044 <title><literal>template</literal>
1045 (mp::filter::Template)</title>
1047 Does nothing at all, merely passing the packet on. (Maybe it
1048 should be called <literal>nop</literal> or
1049 <literal>passthrough</literal>?) This exists not to be used, but
1050 to be copied - to become the skeleton of new filters as they are
1051 written. As with <literal>backend_test</literal>, this is not
1052 intended for civilians.
1057 <title><literal>virt_db</literal>
1058 (mp::filter::VirtualDB)</title>
1060 Performs virtual database selection: based on the name of the
1061 database in the search request, a server is selected, and its
1062 address added to the request in a <literal>VAL_PROXY</literal>
1063 otherInfo packet. It will subsequently be used by a
1064 <literal>z3950_client</literal> filter.
1066 <link linkend="multidb">the extended discussion</link>
1067 of virtual databases and multi-database searching below.
1072 <title><literal>z3950_client</literal>
1073 (mp::filter::Z3950Client)</title>
1075 A partial sink which swallows only Z39.50 packages.
1076 It performs Z39.50 searching and retrieval by proxying the
1077 packages that are passed to it. Init requests are sent to the
1078 address specified in the <literal>VAL_PROXY</literal> otherInfo
1079 attached to the request: this may have been specified by client,
1080 or generated by a <literal>virt_db</literal> filter earlier in
1081 the route. Subsequent requests are sent to the same address,
1082 which is remembered at Init time in a Session object.
1083 HTTP_Request packages and all other forthcoming package types
1084 are passed untouched.
1090 <title><literal>zeerex_explain</literal>
1091 (mp::filter::ZeerexExplain)</title>
1093 This filter acts as a sink for
1094 Z39.50 explain requests, returning a static ZeeReX
1095 Explain XML record from the config section. All other packages
1098 <ulink url="&url.zeerex.explain;">ZeeReX Explain</ulink>
1100 for more information on the correct explain syntax.
1104 This filter is not yet completed.
1113 <section id="future.directions">
1114 <title>Future directions</title>
1116 Some other filters that do not yet exist, but which would be
1117 useful, are briefly described. These may be added in future
1118 releases (or may be created by third parties, as loadable
1124 <term><literal>frontend_cli</literal> (source)</term>
1127 Command-line interface for generating requests.
1132 <term><literal>sru_client</literal> (sink)</term>
1135 SRU/GET and SRU/SOAP searching and retrieval.
1140 <term><literal>opensearch_client</literal> (sink)</term>
1143 A9 OpenSearch searching and retrieval.
1153 <chapter id="configuration">
1154 <title>Configuration: the Metaproxy configuration file format</title>
1158 <title>Introductory notes</title>
1160 If Metaproxy is an interpreter providing operations on packages, then
1161 its configuration file can be thought of as a program for that
1162 interpreter. Configuration is by means of a single XML file, the name
1163 of which is supplied as the sole command-line argument to the
1164 <command>metaproxy</command> program. (See
1165 <xref linkend="reference"/> below for more information on invoking
1170 <section id="overview.xml.structure">
1171 <title>Overview of the config file XML structure</title>
1173 All elements and attributes are in the namespace
1174 <ulink url="http://indexdata.com/metaproxy"/>.
1175 This is most easily achieved by setting the default namespace on
1176 the top-level element, as here:
1179 <metaproxy xmlns="http://indexdata.com/metaproxy" version="1.0">
1182 The top-level element is <metaproxy>. This contains a
1183 <start> element, a <filters> element and a
1184 <routes> element, in that order. <filters> is
1185 optional; the other two are mandatory. All three are
1189 The <start> element is empty, but carries a
1190 <literal>route</literal> attribute, whose value is the name of
1191 route at which to start running - analogous to the name of the
1192 start production in a formal grammar.
1195 If present, <filters> contains zero or more <filter>
1196 elements. Each filter carries a <literal>type</literal> attribute
1197 which specifies what kind of filter is being defined
1198 (<literal>frontend_net</literal>, <literal>log</literal>, etc.)
1199 and contain various elements that provide suitable configuration
1200 for a filter of its type. The filter-specific elements are
1202 <xref linkend="reference"/>.
1203 Filters defined in this part of the file must carry an
1204 <literal>id</literal> attribute so that they can be referenced
1208 <routes> contains one or more <route> elements, each
1209 of which must carry an <literal>id</literal> element. One of the
1210 routes must have the ID value that was specified as the start
1211 route in the <start> element's <literal>route</literal>
1212 attribute. Each route contains zero or more <filter>
1213 elements. These are of two types. They may be empty, but carry a
1214 <literal>refid</literal> attribute whose value is the same as the
1215 <literal>id</literal> of a filter previously defined in the
1216 <filters> section. Alternatively, a route within a filter
1217 may omit the <literal>refid</literal> attribute, but contain
1218 configuration elements similar to those used for filters defined
1219 in the <filters> section. (In other words, each filter in a
1220 route may be included either by reference or by physical
1226 <section id="example.configuration">
1227 <title>An example configuration</title>
1229 The following is a small, but complete, Metaproxy configuration
1230 file (included in the distribution as
1231 <literal>metaproxy/etc/config1.xml</literal>).
1232 This file defines a very simple configuration that simply proxies
1233 to whatever back-end server the client requests, but logs each
1234 request and response. This can be useful for debugging complex
1235 client-server dialogues.
1237 <screen><![CDATA[<?xml version="1.0"?>
1238 <metaproxy xmlns="http://indexdata.com/metaproxy" version="1.0">
1239 <start route="start"/>
1241 <filter id="frontend" type="frontend_net">
1244 <filter id="backend" type="z3950_client">
1249 <filter refid="frontend"/>
1250 <filter type="log"/>
1251 <filter refid="backend"/>
1252 <filter type="bounce"/>
1258 It works by defining a single route, called
1259 <literal>start</literal>, which consists of a sequence of four
1260 filters. The first and last of these are included by reference:
1261 their <literal><filter></literal> elements have
1262 <literal>refid</literal> attributes that refer to filters defined
1263 within the prior <literal><filters></literal> section. The
1264 middle filter is included inline in the route.
1267 The four filters in the route are as follows: first, a
1268 <literal>frontend_net</literal> filter accepts Z39.50 requests
1269 from any host on port 9000; then these requests are passed through
1270 a <literal>log</literal> filter that emits a message for each
1271 request; they are then fed into a <literal>z3950_client</literal>
1272 filter, which forwards all Z39.50 requests to the client-specified
1273 back-end Z39.509 server. Those Z39.50 packages are returned by the
1274 <literal>z3950_client</literal> filter, with the response data
1275 filled by the external Z39.50 server targeted.
1276 All non-Z39.50 packages are passed through to the
1277 <literal>bounce</literal> filter, which definitely bounces
1278 everything, including fish, bananas, cold pyjamas,
1279 mutton, beef and trout packages.
1280 When the response arrives, it is handed
1281 back to the <literal>log</literal> filter, which emits another
1282 message; and then to the <literal>frontend_net</literal> filter,
1283 which returns the response to the client.
1287 <section id="config-file-modularity">
1288 <title>Config file modularity</title>
1290 Metaproxy XML configuration snippets can be reused by other
1291 filters using the <literal>XInclude</literal> standard, as seen in
1292 the <literal>/etc/config-sru-to-z3950.xml</literal> example SRU
1295 <filter id="sru" type="sru_z3950">
1296 <database name="Default">
1297 <xi:include xmlns:xi="http://www.w3.org/2001/XInclude"
1298 href="explain.xml"/>
1305 <section id="config-file-syntax-check">
1306 <title>Config file syntax checking</title>
1308 The distribution contains RelaxNG Compact and XML syntax checking
1309 files, as well as XML Schema files. These are found in the
1312 xml/schema/metaproxy.rnc
1313 xml/schema/metaproxy.rng
1314 xml/schema/metaproxy.xsd
1316 and can be used to verify or debug the XML structure of
1317 configuration files. For example, using the utility
1318 <filename>xmllint</filename>, syntax checking is done like this:
1320 xmllint --noout --schema xml/schema/metaproxy.xsd etc/config-local.xml
1321 xmllint --noout --relaxng xml/schema/metaproxy.rng etc/config-local.xml
1323 (A recent version of <literal>libxml2</literal> is required, as
1324 support for XML Schemas is a relatively recent addition.)
1327 You can of course use any other RelaxNG or XML Schema compliant tool
1335 <chapter id="multidb">
1336 <title>Virtual databases and multi-database searching</title>
1340 <title>Introductory notes</title>
1342 Two of Metaproxy's filters are concerned with multiple-database
1343 operations. Of these, <literal>virt_db</literal> can work alone
1344 to control the routing of searches to one of a number of servers,
1345 while <literal>multi</literal> can work together with
1346 <literal>virt_db</literal> to perform multi-database searching, merging
1347 the results into a unified result-set - ``metasearch in a box''.
1350 The interaction between
1351 these two filters is necessarily complex: it reflects the real,
1352 irreducible complexity of multi-database searching in a protocol such
1353 as Z39.50 that separates initialization from searching, and in
1354 which the database to be searched is not known at initialization
1358 It's possible to use these filters without understanding the
1359 details of their functioning and the interaction between them; the
1360 next two sections of this chapter are ``HOW-TO'' guides for doing
1361 just that. However, debugging complex configurations will require
1362 a deeper understanding, which the last two sections of this
1363 chapters attempt to provide.
1368 <section id="multidb.virt_db">
1369 <title>Virtual databases with the <literal>virt_db</literal> filter</title>
1371 Working alone, the purpose of the
1372 <literal>virt_db</literal>
1373 filter is to route search requests to one of a selection of
1374 back-end databases. In this way, a single Z39.50 endpoint
1375 (running Metaproxy) can provide access to several different
1376 underlying services, including those that would otherwise be
1377 inaccessible due to firewalls. In many useful configurations, the
1378 back-end databases are local to the Metaproxy installation, but
1379 the software does not enforce this, and any valid Z39.50 servers
1380 may be used as back-ends.
1383 For example, a <literal>virt_db</literal>
1384 filter could be set up so that searches in the virtual database
1385 ``lc'' are forwarded to the Library of Congress bibliographic
1386 catalogue server, and searches in the virtual database ``marc''
1387 are forwarded to the toy database of MARC records that Index Data
1388 hosts for testing purposes. A <literal>virt_db</literal>
1389 configuration to make this switch would look like this:
1391 <screen><![CDATA[<filter type="virt_db">
1393 <database>lc</database>
1394 <target>z3950.loc.gov:7090/voyager</target>
1397 <database>marc</database>
1398 <target>indexdata.com/marc</target>
1400 </filter>]]></screen>
1402 As well as being useful in it own right, this filter also provides
1403 the foundation for multi-database searching.
1408 <section id="multidb.multi">
1409 <title>Multi-database search with the <literal>multi</literal> filter</title>
1411 To arrange for Metaproxy to broadcast searches to multiple back-end
1412 servers, the configuration needs to include two components: a
1413 <literal>virt_db</literal>
1414 filter that specifies multiple
1415 <literal><target></literal>
1416 elements, and a subsequent
1417 <literal>multi</literal>
1418 filter. Here, for example, is a complete configuration that
1419 broadcasts searches to both the Library of Congress catalogue and
1420 Index Data's tiny testing database of MARC records:
1422 <screen><![CDATA[<?xml version="1.0"?>
1423 <metaproxy xmlns="http://indexdata.com/metaproxy" version="1.0">
1424 <start route="start"/>
1427 <filter type="frontend_net">
1428 <threads>10</threads>
1431 <filter type="virt_db">
1433 <database>lc</database>
1434 <target>z3950.loc.gov:7090/voyager</target>
1437 <database>marc</database>
1438 <target>indexdata.com/marc</target>
1441 <database>all</database>
1442 <target>z3950.loc.gov:7090/voyager</target>
1443 <target>indexdata.com/marc</target>
1446 <filter type="multi"/>
1447 <filter type="z3950_client">
1448 <timeout>30</timeout>
1450 <filter type="bounce"/>
1453 </metaproxy>]]></screen>
1456 <literal>virt_db</literal>
1457 filter that specifies multiple
1458 <literal><target></literal>
1459 elements but without a subsequent
1460 <literal>multi</literal>
1461 filter yields surprising and undesirable results, as will be
1462 described below. Don't do that.)
1465 Metaproxy can be invoked with this configuration as follows:
1467 <screen>../src/metaproxy --config config-simple-multi.xml</screen>
1469 And thereafter, Z39.50 clients can connect to the running server
1470 (on port 9000, as specified in the configuration) and search in
1471 any of the databases
1472 <literal>lc</literal> (the Library of Congress catalogue),
1473 <literal>marc</literal> (Index Data's test database of MARC records)
1475 <literal>all</literal> (both of these). As an example, a session
1476 using the YAZ command-line client <literal>yaz-client</literal> is
1477 here included (edited for brevity and clarity):
1479 <screen><![CDATA[$ yaz-client @:9000
1483 Search was a success.
1484 Number of hits: 10000, setno 1
1488 Search was a success.
1489 Number of hits: 10, setno 3
1493 Search was a success.
1494 Number of hits: 10010, setno 4
1497 [marc]Record type: USmarc
1500 005 00000000000000.0
1501 008 910710c19910701nju 00010 eng
1505 100 10 $a Jack Collins
1506 245 10 $a How to program a computer
1512 [VOYAGER]Record type: USmarc
1514 005 20041229102447.0
1515 008 030910s2004 caua 000 0 eng
1516 035 $a (DLC) 2003112666
1517 906 $a 7 $b cbc $c orignew $d 4 $e epcn $f 20 $g y-gencatlg
1518 925 0 $a acquire $b 1 shelf copy $x policy default
1519 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.
1522 040 $a DLC $c DLC $d DLC
1523 050 00 $a MLCM 2004/03312 (G)
1524 245 10 $a 007, everything or nothing : $b Prima's official strategy guide / $c created by Kaizen Media Group.
1525 246 3 $a Double-O-seven, everything or nothing
1526 246 30 $a Prima's official strategy guide
1527 260 $a Roseville, CA : $b Prima Games, $c c2004.
1528 300 $a 161 p. : $b col. ill. ; $c 28 cm.
1529 500 $a "Platforms: Nintendo GameCube, Macintosh, PC, PlayStation 2 computer entertainment system, Xbox"--P. [4] of cover.
1530 650 0 $a Video games.
1531 710 2 $a Kaizen Media Group.
1532 856 42 $3 Publisher description $u http://www.loc.gov/catdir/description/random052/2003112666.html
1537 As can be seen, the first record in the result set is from the
1538 Index Data test database, and the second from the Library of
1539 Congress database. The result-set continues alternating records
1540 round-robin style until the point where one of the databases'
1541 records are exhausted.
1544 This example uses only two back-end databases; more may be used.
1545 There is no limitation imposed on the number of databases that may
1546 be metasearched in this way: issues of resource usage and
1547 administrative complexity dictate the practical limits.
1550 What happens when one of the databases doesn't respond? By default,
1551 the entire multi-database search fails, and the appropriate
1552 diagnostic is returned to the client. This is usually appropriate
1553 during development, when technicians need maximum information, but
1554 can be inconvenient in deployment, when users typically don't want
1555 to be bothered with problems of this kind and prefer just to get
1556 the records from the databases that are available. To obtain this
1557 latter behavior add an empty
1558 <literal><hideunavailable></literal>
1560 <literal>multi</literal> filter:
1562 <screen><![CDATA[ <filter type="multi">
1564 </filter>]]></screen>
1566 Under this regime, an error is reported to the client only if
1567 <emphasis>all</emphasis> the databases in a multi-database search
1573 <section id="multidb.what">
1574 <title>What's going on?</title>
1576 <title>Lark's vomit</title>
1578 This section goes into a level of technical detail that is
1579 probably not necessary in order to configure and use Metaproxy.
1580 It is provided only for those who like to know how things work.
1581 You should feel free to skip on to the next section if this one
1582 doesn't seem like fun.
1586 Hold on tight - this may get a little hairy.
1589 In the general course of things, a Z39.50 Init request may carry
1590 with it an otherInfo packet of type <literal>VAL_PROXY</literal>,
1591 whose value indicates the address of a Z39.50 server to which the
1592 ultimate connection is to be made. (This otherInfo packet is
1593 supported by YAZ-based Z39.50 clients and servers, but has not yet
1594 been ratified by the Maintenance Agency and so is not widely used
1595 in non-Index Data software. We're working on it.)
1596 The <literal>VAL_PROXY</literal> packet functions
1597 analogously to the absoluteURI-style Request-URI used with the GET
1598 method when a web browser asks a proxy to forward its request: see
1600 <ulink url="http://www.w3.org/Protocols/rfc2616/rfc2616-sec5.html#sec5.1.2"
1601 >Request-URI</ulink>
1603 <ulink url="http://www.w3.org/Protocols/rfc2616/rfc2616.html"
1604 >the HTTP 1.1 specification</ulink>.
1607 Within Metaproxy, Search requests that are part of the same
1608 session as an Init request that carries a
1609 <literal>VAL_PROXY</literal> otherInfo are also annotated with the
1610 same information. The role of the <literal>virt_db</literal>
1611 filter is to rewrite this otherInfo packet dependent on the
1612 virtual database that the client wants to search.
1615 When Metaproxy receives a Z39.50 Init request from a client, it
1616 doesn't immediately forward that request to the back-end server.
1617 Why not? Because it doesn't know <emphasis>which</emphasis>
1618 back-end server to forward it to until the client sends a Search
1619 request that specifies the database that it wants to search in.
1620 Instead, it just treasures the Init request up in its heart; and,
1621 later, the first time the client does a search on one of the
1622 specified virtual databases, a connection is forged to the
1623 appropriate server and the Init request is forwarded to it. If,
1624 later in the session, the same client searches in a different
1625 virtual database, then a connection is forged to the server that
1626 hosts it, and the same cached Init request is forwarded there,
1630 All of this clever Init-delaying is done by the
1631 <literal>frontend_net</literal> filter. The
1632 <literal>virt_db</literal> filter knows nothing about it; in
1633 fact, because the Init request that is received from the client
1634 doesn't get forwarded until a Search request is received, the
1635 <literal>virt_db</literal> filter (and the
1636 <literal>z3950_client</literal> filter behind it) doesn't even get
1637 invoked at Init time. The <emphasis>only</emphasis> thing that a
1638 <literal>virt_db</literal> filter ever does is rewrite the
1639 <literal>VAL_PROXY</literal> otherInfo in the requests that pass
1643 It is possible for a <literal>virt_db</literal> filter to contain
1645 <literal><target></literal>
1646 elements. What does this mean? Only that the filter will add
1647 multiple <literal>VAL_PROXY</literal> otherInfo packets to the
1648 Search requests that pass through it. That's because the virtual
1649 DB filter is dumb, and does exactly what it's told - no more, no
1651 If a Search request with multiple <literal>VAL_PROXY</literal>
1652 otherInfo packets reaches a <literal>z3950_client</literal>
1653 filter, this is an error. That filter doesn't know how to deal
1654 with multiple targets, so it will either just pick one and search
1655 in it, or (better) fail with an error message.
1658 The <literal>multi</literal> filter comes to the rescue! This is
1659 the only filter that knows how to deal with multiple
1660 <literal>VAL_PROXY</literal> otherInfo packets, and it does so by
1661 making multiple copies of the entire Search request: one for each
1662 <literal>VAL_PROXY</literal>. Each of these new copies is then
1663 passed down through the remaining filters in the route. (The
1664 copies are handled in parallel though the
1665 spawning of new threads.) Since the copies each have only one
1666 <literal>VAL_PROXY</literal> otherInfo, they can be handled by the
1667 <literal>z3950_client</literal> filter, which happily deals with
1668 each one individually. When the results of the individual
1669 searches come back up to the <literal>multi</literal> filter, it
1670 merges them into a single Search response, which is what
1671 eventually makes it back to the client.
1676 <section id="multidb.picture">
1677 <title>A picture is worth a thousand words (but only five hundred on 64-bit architectures)</title>
1681 <imagedata fileref="multi.pdf" format="PDF" scale="50"/>
1684 <imagedata fileref="multi.png" format="PNG"/>
1687 <!-- Fall back if none of the images can be used -->
1689 [Here there should be a diagram showing the progress of
1690 packages through the filters during a simple virtual-database
1691 search and a multi-database search, but is seems that your
1692 tool chain has not been able to include the diagram in this
1696 <!-- ### This used to work with an older version of DocBook
1698 <para>Caption: progress of packages through filters.</para>
1701 </inlinemediaobject>
1707 <chapter id="sru-server">
1708 <title>Combined SRU webservice and Z39.50 server configuration</title>
1710 Metaproxy can act as
1711 <ulink url="&url.sru;">SRU</ulink> and
1712 <ulink url="&url.srw;">SRW</ulink>
1713 web service server, which translates web service requests to
1714 <ulink url="&url.z39.50;">ANSI/NISO Z39.50</ulink> packages and
1715 sends them off to common available targets.
1718 A typical setup for this operation needs a filter route including the
1722 <table id="sru-server-table-config" frame="top">
1723 <title>SRU/Z39.50 Server Filter Route Configuration</title>
1727 <entry>Filter</entry>
1728 <entry>Importance</entry>
1729 <entry>Purpose</entry>
1735 <entry><literal>frontend_net</literal></entry>
1736 <entry>required</entry>
1737 <entry>Accepting HTTP connections and passing them to following
1738 filters. Since this filter also accepts Z39.50 connections, the
1739 server works as SRU and Z39.50 server on the same port.</entry>
1742 <entry><literal>sru_z3950</literal></entry>
1743 <entry>required</entry>
1744 <entry>Accepting SRU GET/POST/SOAP explain and
1745 searchRetrieve requests for the the configured databases.
1746 Explain requests are directly served from the static XML configuration.
1747 SearchRetrieve requests are
1748 transformed to Z39.50 search and present packages.
1749 All other HTTP and Z39.50 packages are passed unaltered.</entry>
1752 <entry><literal>http_file</literal></entry>
1753 <entry>optional</entry>
1754 <entry>Serving HTTP requests from the filesystem. This is only
1755 needed if the server should serve XSLT stylesheets, static HTML
1756 files or Java Script for thin browser based clients.
1757 Z39.50 packages are passed unaltered.</entry>
1760 <entry><literal>cql_rpn</literal></entry>
1761 <entry>required</entry>
1762 <entry>Usually, Z39.50 servers do not talk CQL, hence the
1763 translation of the CQL query language to RPN is mandatory in
1764 most cases. Affects only Z39.50 search packages.</entry>
1767 <entry><literal>record_transform</literal></entry>
1768 <entry>optional</entry>
1769 <entry>Some Z39.50 backend targets can not present XML record
1770 syntaxes in common wanted element sets. using this filter, one
1771 can transform binary MARC records to MARCXML records, and
1772 further transform those to any needed XML schema/format by XSLT
1773 transformations. Changes only Z39.50 present packages.</entry>
1776 <entry><literal>session_shared</literal></entry>
1777 <entry>optional</entry>
1778 <entry>The stateless nature of web services requires frequent
1779 re-searching of the same targets for display of paged result set
1780 records. This might be an unacceptable burden for the accessed
1781 backend Z39.50 targets, and this mosule can be added for
1782 efficient backend target resource pooling.</entry>
1785 <entry><literal>z3950_client</literal></entry>
1786 <entry>required</entry>
1787 <entry>Finally, a Z39.50 package sink is needed in the filter
1788 chain to provide the response packages. The Z39.50 client module
1789 is used to access external targets over the network, but any
1790 coming local Z39.50 package sink could be used instead of.</entry>
1793 <entry><literal>bounce</literal></entry>
1794 <entry>required</entry>
1795 <entry>Any Metaproxy package arriving here did not do so by
1796 purpose, and is bounced back with connection closure. this
1797 prevents inifinite package hanging inside the SRU server.</entry>
1803 A typical minimal example <ulink url="&url.sru;">SRU</ulink> and
1804 <ulink url="&url.srw;">SRW</ulink> server configuration file is found
1805 in the tarball distribution at
1806 <literal>etc/config-sru-to-z3950.xml</literal>.
1809 Off course, any other metaproxy modules can be integrated into a
1810 SRU server solution, including, but not limited to, load balancing,
1811 multiple target querying
1812 (see <xref linkend="multidb"/>), and complex RPN query rewrites.
1819 <chapter id="extensions">
1820 <title>Writing extensions for Metaproxy</title>
1821 <para>### To be written</para>
1827 <chapter id="classes">
1828 <title>Classes in the Metaproxy source code</title>
1832 <title>Introductory notes</title>
1834 <emphasis>Stop! Do not read this!</emphasis>
1835 You won't enjoy it at all. You should just skip ahead to
1836 <xref linkend="reference"/>,
1838 <!-- The remainder of this paragraph is lifted verbatim from
1839 Douglas Adams' _Hitch Hiker's Guide to the Galaxy_, chapter 8 -->
1840 you things you really need to know, like the fact that the
1841 fabulously beautiful planet Bethselamin is now so worried about
1842 the cumulative erosion by ten billion visiting tourists a year
1843 that any net imbalance between the amount you eat and the amount
1844 you excrete whilst on the planet is surgically removed from your
1845 bodyweight when you leave: so every time you go to the lavatory it
1846 is vitally important to get a receipt.
1849 This chapter contains documentation of the Metaproxy source code, and is
1850 of interest only to maintainers and developers. If you need to
1851 change Metaproxy's behavior or write a new filter, then you will most
1852 likely find this chapter helpful. Otherwise it's a waste of your
1853 good time. Seriously: go and watch a film or something.
1854 <citetitle>This is Spinal Tap</citetitle> is particularly good.
1857 Still here? OK, let's continue.
1860 In general, classes seem to be named big-endianly, so that
1861 <literal>FactoryFilter</literal> is not a filter that filters
1862 factories, but a factory that produces filters; and
1863 <literal>FactoryStatic</literal> is a factory for the statically
1864 registered filters (as opposed to those that are dynamically
1869 <section id="individual.classes">
1870 <title>Individual classes</title>
1872 The classes making up the Metaproxy application are here listed by
1873 class-name, with the names of the source files that define them in
1878 <title><literal>mp::FactoryFilter</literal>
1879 (<filename>factory_filter.cpp</filename>)</title>
1881 A factory class that exists primarily to provide the
1882 <literal>create()</literal> method, which takes the name of a
1883 filter class as its argument and returns a new filter of that
1884 type. To enable this, the factory must first be populated by
1885 calling <literal>add_creator()</literal> for static filters (this
1886 is done by the <literal>FactoryStatic</literal> class, see below)
1887 and <literal>add_creator_dyn()</literal> for filters loaded
1893 <title><literal>mp::FactoryStatic</literal>
1894 (<filename>factory_static.cpp</filename>)</title>
1896 A subclass of <literal>FactoryFilter</literal> which is
1897 responsible for registering all the statically defined filter
1898 types. It does this by knowing about all those filters'
1899 structures, which are listed in its constructor. Merely
1900 instantiating this class registers all the static classes. It is
1901 for the benefit of this class that <literal>struct
1902 metaproxy_1_filter_struct</literal> exists, and that all the filter
1903 classes provide a static object of that type.
1908 <title><literal>mp::filter::Base</literal>
1909 (<filename>filter.cpp</filename>)</title>
1911 The virtual base class of all filters. The filter API is, on the
1912 surface at least, extremely simple: two methods.
1913 <literal>configure()</literal> is passed an XML DOM tree representing
1914 that part of the configuration file that pertains to this filter
1915 instance, and is expected to walk that tree extracting relevant
1916 information. And <literal>process()</literal> processes a
1917 package (see below). That surface simplicity is a bit
1918 misleading, as <literal>process()</literal> needs to know a lot
1919 about the <literal>Package</literal> class in order to do
1925 <title><literal>mp::filter::AuthSimple</literal>,
1926 <literal>Backend_test</literal>, etc.
1927 (<filename>filter_auth_simple.cpp</filename>,
1928 <filename>filter_backend_test.cpp</filename>, etc.)</title>
1930 Individual filters. Each of these is implemented by a header and
1931 a source file, named <filename>filter_*.hpp</filename> and
1932 <filename>filter_*.cpp</filename> respectively. All the header
1933 files should be pretty much identical, in that they declare the
1934 class, including a private <literal>Rep</literal> class and a
1935 member pointer to it, and the two public methods.
1938 The source file for each filter needs to supply:
1943 A definition of the private <literal>Rep</literal> class.
1948 Some boilerplate constructors and destructors.
1953 A <literal>configure()</literal> method that uses the
1954 appropriate XML fragment.
1959 Most important, the <literal>process()</literal> method that
1960 does all the actual work.
1967 <title><literal>mp::Package</literal>
1968 (<filename>package.cpp</filename>)</title>
1970 Represents a package on its way through the series of filters
1971 that make up a route. This is essentially a Z39.50 or SRU APDU
1972 together with information about where it came from, which is
1973 modified as it passes through the various filters.
1978 <title><literal>mp::Pipe</literal>
1979 (<filename>pipe.cpp</filename>)</title>
1981 This class provides a compatibility layer so that we have an IPC
1982 mechanism that works the same under Unix and Windows. It's not
1983 particularly exciting.
1988 <title><literal>mp::RouterChain</literal>
1989 (<filename>router_chain.cpp</filename>)</title>
1996 <title><literal>mp::RouterFleXML</literal>
1997 (<filename>router_flexml.cpp</filename>)</title>
2004 <title><literal>mp::Session</literal>
2005 (<filename>session.cpp</filename>)</title>
2012 <title><literal>mp::ThreadPoolSocketObserver</literal>
2013 (<filename>thread_pool_observer.cpp</filename>)</title>
2020 <title><literal>mp::util</literal>
2021 (<filename>util.cpp</filename>)</title>
2023 A namespace of various small utility functions and classes,
2024 collected together for convenience. Most importantly, includes
2025 the <literal>mp::util::odr</literal> class, a wrapper for YAZ's
2031 <title><literal>mp::xml</literal>
2032 (<filename>xmlutil.cpp</filename>)</title>
2034 A namespace of various XML utility functions and classes,
2035 collected together for convenience.
2041 <section id="other.source.files">
2042 <title>Other Source Files</title>
2044 In addition to the Metaproxy source files that define the classes
2045 described above, there are a few additional files which are
2046 briefly described here:
2050 <term><literal>metaproxy_prog.cpp</literal></term>
2053 The main function of the <command>metaproxy</command> program.
2058 <term><literal>ex_router_flexml.cpp</literal></term>
2061 Identical to <literal>metaproxy_prog.cpp</literal>: it's not clear why.
2066 <term><literal>test_*.cpp</literal></term>
2069 Unit-tests for various modules.
2075 ### Still to be described:
2076 <literal>ex_filter_frontend_net.cpp</literal>,
2077 <literal>filter_dl.cpp</literal>,
2078 <literal>plainfile.cpp</literal>,
2079 <literal>tstdl.cpp</literal>.
2085 <reference id="reference">
2086 <title>Reference</title>
2089 The material in this chapter is drawn directly from the individual
2090 manual entries. In particular, the Metaproxy invocation section is
2091 available using <command>man metaproxy</command>, and the section
2092 on each individual filter is available using the name of the filter
2093 as the argument to the <command>man</command> command.
2100 <!-- Keep this comment at the end of the file
2105 sgml-minimize-attributes:nil
2106 sgml-always-quote-attributes:t
2109 sgml-parent-document: nil
2110 sgml-local-catalogs: nil
2111 sgml-namecase-general:t