1 <chapter id="record-model-domxml">
2 <!-- $Id: recordmodel-domxml.xml,v 1.10 2007-03-01 11:21:20 marc Exp $ -->
3 <title>&dom; &xml; Record Model and Filter Module</title>
6 The record model described in this chapter applies to the fundamental,
8 record type <literal>&dom;</literal>, introduced in
9 <xref linkend="componentmodulesdom"/>. The &dom; &xml; record model
10 is experimental, and it's inner workings might change in future
11 releases of the &zebra; Information Server.
16 <section id="record-model-domxml-filter">
17 <title>&dom; Record Filter Architecture</title>
20 The &dom; &xml; filter uses a standard &dom; &xml; structure as
21 internal data model, and can therefore parse, index, and display
22 any &xml; document type. It is well suited to work on
23 standardized &xml;-based formats such as Dublin Core, MODS, METS,
24 MARCXML, OAI-PMH, RSS, and performs equally well on any other
25 non-standard &xml; format.
28 A parser for binary &marc; records based on the ISO2709 library
29 standard is provided, it transforms these to the internal
30 &marcxml; &dom; representation. Other binary document parsers
31 are planned to follow.
35 The &dom; filter architecture consists of four
36 different pipelines, each being a chain of arbitrarily many successive
37 &xslt; transformations of the internal &dom; &xml;
38 representations of documents.
41 <figure id="record-model-domxml-architecture-fig">
42 <title>&dom; &xml; filter architecture</title>
45 <imagedata fileref="domfilter.pdf" format="PDF" scale="50"/>
48 <imagedata fileref="domfilter.png" format="PNG"/>
51 <!-- Fall back if none of the images can be used -->
53 [Here there should be a diagram showing the &dom; &xml;
54 filter architecture, but is seems that your
55 tool chain has not been able to include the diagram in this
63 <table id="record-model-domxml-architecture-table" frame="top">
64 <title>&dom; &xml; filter pipelines overview</title>
70 <entry>Description</entry>
78 <entry><literal>input</literal></entry>
80 <entry>input parsing and initial
81 transformations to common &xml; format</entry>
82 <entry>Input raw &xml; record buffers, &xml; streams and
83 binary &marc; buffers</entry>
84 <entry>Common &xml; &dom;</entry>
87 <entry><literal>extract</literal></entry>
89 <entry>indexing term extraction
90 transformations</entry>
91 <entry>Common &xml; &dom;</entry>
92 <entry>Indexing &xml; &dom;</entry>
95 <entry><literal>store</literal></entry>
97 <entry> transformations before internal document
99 <entry>Common &xml; &dom;</entry>
100 <entry>Storage &xml; &dom;</entry>
103 <entry><literal>retrieve</literal></entry>
105 <entry>multiple document retrieve transformations from
106 storage to different output
107 formats are possible</entry>
108 <entry>Storage &xml; &dom;</entry>
109 <entry>Output &xml; syntax in requested formats</entry>
116 The &dom; &xml; filter pipelines use &xslt; (and if supported on
117 your platform, even &exslt;), it brings thus full &xpath;
118 support to the indexing, storage and display rules of not only
119 &xml; documents, but also binary &marc; records.
124 <section id="record-model-domxml-pipeline">
125 <title>&dom; &xml; filter pipeline configuration</title>
128 The experimental, loadable &dom; &xml;/&xslt; filter module
129 <literal>mod-dom.so</literal>
130 is invoked by the <filename>zebra.cfg</filename> configuration statement
132 recordtype.xml: dom.db/filter_dom_conf.xml
134 In this example the &dom; &xml; filter is configured to work
135 on all data files with suffix
136 <filename>*.xml</filename>, where the configuration file is found in the
137 path <filename>db/filter_dom_conf.xml</filename>.
140 <para>The &dom; &xslt; filter configuration file must be
141 valid &xml;. It might look like this:
144 <?xml version="1.0" encoding="UTF8"?>
145 <dom xmlns="http://indexdata.com/zebra-2.0">
147 <xmlreader level="1"/>
148 <!-- <marc inputcharset="marc-8"/> -->
151 <xslt stylesheet="common2index.xsl"/>
154 <xslt stylesheet="common2store.xsl"/>
157 <xslt stylesheet="store2dc.xsl"/>
159 <retrieve name="mods">
160 <xslt stylesheet="store2mods.xsl"/>
167 The root &xml; element <literal><dom></literal> and all other &dom;
168 &xml; filter elements are residing in the namespace
169 <literal>xmlns="http://indexdata.dk/zebra-2.0"</literal>.
172 All pipeline definition elements - i.e. the
173 <literal><input></literal>,
174 <literal><extract></literal>,
175 <literal><store></literal>, and
176 <literal><retrieve></literal> elements - are optional.
177 Missing pipeline definitions are just interpreted
178 do-nothing identity pipelines.
181 All pipeline definition elements may contain zero or more
182 <literal><![CDATA[<xslt stylesheet="path/file.xsl"/>]]></literal>
183 &xslt; transformation instructions, which are performed
184 sequentially from top to bottom.
185 The paths in the <literal>stylesheet</literal> attributes
186 are relative to zebras working directory, or absolute to the file
191 <section id="record-model-domxml-pipeline-input">
192 <title>Input pipeline</title>
194 The <literal><input></literal> pipeline definition element
195 may contain either one &xml; Reader definition
196 <literal><![CDATA[<xmlreader level="1"/>]]></literal>, used to split
197 an &xml; collection input stream into individual &xml; &dom;
198 documents at the prescribed element level,
201 <literal><![CDATA[<marc inputcharset="marc-8"/>]]></literal>, which defines
202 a conversion to &marcxml; format &dom; trees. The allowed values
203 of the <literal>inputcharset</literal> attribute depend on your
204 local <productname>iconv</productname> set-up.
207 Both input parsers deliver individual &dom; &xml; documents to the
208 following chain of zero or more
209 <literal><![CDATA[<xslt stylesheet="path/file.xsl"/>]]></literal>
210 &xslt; transformations. At the end of this pipeline, the documents
211 are in the common format, used to feed both the
212 <literal><extract></literal> and
213 <literal><store></literal> pipelines.
217 <section id="record-model-domxml-pipeline-extract">
218 <title>Extract pipeline</title>
220 The <literal><extract></literal> pipeline takes documents
221 from any common &dom; &xml; format to the &zebra; specific
222 indexing &dom; &xml; format.
223 It may consist of zero ore more
224 <literal><![CDATA[<xslt stylesheet="path/file.xsl"/>]]></literal>
225 &xslt; transformations, and the outcome is handled to the
226 &zebra; core to drive the process of building the inverted
228 <xref linkend="record-model-domxml-canonical-index"/> for
233 <section id="record-model-domxml-pipeline-store">
234 <title>Store pipeline</title>
235 The <literal><store></literal> pipeline takes documents
236 from any common &dom; &xml; format to the &zebra; specific
237 storage &dom; &xml; format.
238 It may consist of zero ore more
239 <literal><![CDATA[<xslt stylesheet="path/file.xsl"/>]]></literal>
240 &xslt; transformations, and the outcome is handled to the
241 &zebra; core for deposition into the internal storage system.
244 <section id="record-model-domxml-pipeline-retrieve">
245 <title>Retrieve pipeline</title>
247 Finally, there may be one or more
248 <literal><retrieve></literal> pipeline definitions, each
249 of them again consisting of zero or more
250 <literal><![CDATA[<xslt stylesheet="path/file.xsl"/>]]></literal>
251 &xslt; transformations. These are used for document
252 presentation after search, and take the internal storage &dom;
253 &xml; to the requested output formats during record present
257 The possible multiple
258 <literal><retrieve></literal> pipeline definitions
259 are distinguished by their unique <literal>name</literal>
260 attributes, these are the literal <literal>schema</literal> or
261 <literal>element set</literal> names used in
262 <ulink url="http://www.loc.gov/standards/sru/srw/">&srw;</ulink>,
263 <ulink url="&url.sru;">&sru;</ulink> and
264 &z3950; protocol queries.
269 <section id="record-model-domxml-canonical-index">
270 <title>Canonical Indexing Format</title>
273 &dom; &xml; indexing comes in two flavors: pure
274 processing-instruction governed plain &xml; documents, and - very
275 similar to the Alvis filter indexing format - &xml; documents
276 containing &xml; <literal><record></literal> and
277 <literal><index></literal> instructions from the magic
278 namespace <literal>xmlns:z="http://indexdata.dk/zebra-2.0"</literal>.
281 <section id="record-model-domxml-canonical-index-pi">
282 <title>Processing-instruction governed indexing format</title>
284 <para>The output of the processing instruction driven
285 indexing &xslt; stylesheets must contain
286 processing instructions named
287 <literal>zebra-2.0</literal>.
288 The output of the &xslt; indexing transformation is then
289 parsed using &dom; methods, and the contained instructions are
290 performed on the <emphasis>elements and their
291 subtrees directly following the processing instructions</emphasis>.
294 For example, the output of the command
296 xsltproc dom-index-pi.xsl marc-one.xml
298 might look like this:
301 <?xml version="1.0" encoding="UTF-8"?>
302 <?zebra-2.0 record id=11224466 rank=42?>
304 <?zebra-2.0 index control:0?>
305 <control>11224466</control>
306 <?zebra-2.0 index any:w title:w title:p title:s?>
307 <title>How to program a computer</title>
314 <section id="record-model-domxml-canonical-index-element">
315 <title>Magic element governed indexing format</title>
317 <para>The output of the indexing &xslt; stylesheets must contain
318 certain elements in the magic
319 <literal>xmlns:z="http://indexdata.dk/zebra-2.0"</literal>
320 namespace. The output of the &xslt; indexing transformation is then
321 parsed using &dom; methods, and the contained instructions are
322 performed on the <emphasis>magic elements and their
326 For example, the output of the command
328 xsltproc dom-index-element.xsl marc-one.xml
330 might look like this:
333 <?xml version="1.0" encoding="UTF-8"?>
334 <z:record xmlns:z="http://indexdata.com/zebra-2.0"
335 z:id="11224466" z:rank="42">
336 <z:index name="control:0">11224466</z:index>
337 <z:index name="any:w title:w title:p title:s">
338 How to program a computer</z:index>
346 <section id="record-model-domxml-canonical-index-semantics">
347 <title>Semantics of the indexing formats</title>
350 Both indexing formats are defined with equal semantics and
354 <para>&zebra; specific instructions are either
355 processing instructions named
356 <literal>zebra-2.0</literal> or
357 elements contained in the namespace
358 <literal>xmlns:z="http://indexdata.dk/zebra-2.0"</literal>.
362 <para>There must be exactly one <literal>record</literal>
363 instruction, which sets the scope for the following,
364 possibly nested <literal>index</literal> instructions.
368 <para>The unique <literal>record</literal> instruction
369 may have additional attributes <literal>id</literal> and
370 <literal>rank</literal>, where the value of the opaque ID
371 may be any string not containing the whitespace character
372 <literal>' '</literal>, and the rank value must be a
373 non-negative integer. See
374 <xref linkend="administration-ranking"/>
378 <para> Multiple and possible nested <literal>index</literal>
379 instructions must contain at least one
380 <literal>indexname:indextype</literal>
381 pair, and may contain multiple such pairs separated by the
382 whitespace character <literal>' '</literal>. In each index
383 pair, the name and the type of the index is separated by a
384 colon character <literal>':'</literal>.
389 Any index name consisting of ASCII letters, and following the
390 standard &zebra; rules will do, see
391 <xref linkend="querymodel-pqf-apt-mapping-accesspoint"/>.
396 Index types are restricted to the values defined in
397 the standard configuration
398 file <filename>default.idx</filename>, see
399 <xref linkend="querymodel-bib1"/> and
400 <xref linkend="fields-and-charsets"/> for details.
405 &dom; input documents which are not resulting in both one
407 <literal>record</literal> instruction and one or more valid
408 <literal>index</literal> instructions can not be searched and
410 invalid document processing is aborted, and any content of
411 the <literal><extract></literal> and
412 <literal><store></literal> pipelines is discarted.
413 A warning is issued in the logs.
420 <para>The examples work as follows:
421 From the original &xml; file
422 <literal>marc-one.xml</literal> (or from the &xml; record &dom; of the
423 same form coming from an <literal><input></literal>
426 pipeline <literal><extract></literal>
427 produces an indexing &xml; record, which is defined by
428 the <literal>record</literal> instruction
429 &zebra; uses the content of
430 <literal>z:id="11224466"</literal>
432 <literal>id=11224466</literal>
434 record ID, and - in case static ranking is set - the content of
435 <literal>rank=42</literal>
437 <literal>z:rank="42"</literal>
442 <para>In these examples, the following literal indexes are constructed:
450 where the indexing type is defined after the
451 literal <literal>':'</literal> character.
452 Any value from the standard configuration
453 file <filename>default.idx</filename> will do.
455 <literal>text()</literal> node content recursively contained
456 inside the <literal><z:index></literal> element, or any
457 element following a <literal>index</literal> processing instruction,
458 will be filtered through the
459 appropriate char map for character normalization, and will be
460 inserted in the named indexes.
463 Finally, this example configuration can be queried using &pqf;
464 queries, either transported by &z3950;, (here using a yaz-client)
467 Z> open localhost:9999
471 Z> find @attr 1=control @attr 4=3 11224466
472 Z> scan @attr 1=control @attr 4=3 ""
474 Z> find @attr 1=title program
475 Z> scan @attr 1=title ""
477 Z> find @attr 1=title @attr 4=2 "How to program a computer"
478 Z> scan @attr 1=title @attr 4=2 ""
482 extensions <literal>x-pquery</literal> and
483 <literal>x-pScanClause</literal> to
487 http://localhost:9999/?version=1.1&operation=searchRetrieve&x-pquery=@attr 1=title program
488 http://localhost:9999/?version=1.1&operation=scan&x-pScanClause=@attr 1=title ""
491 See <xref linkend="zebrasrv-sru"/> for more information on &sru;/&srw;
492 configuration, and <xref linkend="gfs-config"/> or the &yaz;
493 <ulink url="&url.yaz.cql;">&cql; section</ulink>
494 for the details or the &yaz; frontend server.
497 Notice that there are no <filename>*.abs</filename>,
498 <filename>*.est</filename>, <filename>*.map</filename>, or other &grs1;
499 filter configuration files involves in this process, and that the
500 literal index names are used during search and retrieval.
503 In case that we want to support the usual
504 <literal>bib-1</literal> &z3950; numeric access points, it is a
505 good idea to choose string index names defined in the default
506 configuration file <filename>tab/bib1.att</filename>, see
507 <xref linkend="attset-files"/>
516 <section id="record-model-domxml-conf">
517 <title>&dom; Record Model Configuration</title>
520 <section id="record-model-domxml-index">
521 <title>&dom; Indexing Configuration</title>
523 As mentioned above, there can be only one indexing pipeline,
524 and configuration of the indexing process is a synonym
525 of writing an &xslt; stylesheet which produces &xml; output containing the
526 magic processing instructions or elements discussed in
527 <xref linkend="record-model-domxml-canonical-index"/>.
528 Obviously, there are million of different ways to accomplish this
529 task, and some comments and code snippets are in order to
533 Stylesheets can be written in the <emphasis>pull</emphasis> or
534 the <emphasis>push</emphasis> style: <emphasis>pull</emphasis>
535 means that the output &xml; structure is taken as starting point of
536 the internal structure of the &xslt; stylesheet, and portions of
537 the input &xml; are <emphasis>pulled</emphasis> out and inserted
538 into the right spots of the output &xml; structure.
540 side, <emphasis>push</emphasis> &xslt; stylesheets are recursively
541 calling their template definitions, a process which is commanded
542 by the input &xml; structure, and is triggered to produce
544 whenever some special conditions in the input stylesheets are
545 met. The <emphasis>pull</emphasis> type is well-suited for input
546 &xml; with strong and well-defined structure and semantics, like the
547 following &oai; indexing example, whereas the
548 <emphasis>push</emphasis> type might be the only possible way to
549 sort out deeply recursive input &xml; formats.
552 A <emphasis>pull</emphasis> stylesheet example used to index
553 &oai; harvested records could use some of the following template
557 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
558 xmlns:z="http://indexdata.dk/zebra-2.0"
559 xmlns:oai="http://www.openarchives.org/&oai;/2.0/"
560 xmlns:oai_dc="http://www.openarchives.org/&oai;/2.0/oai_dc/"
561 xmlns:dc="http://purl.org/dc/elements/1.1/"
564 <!-- Example pull and magic element style Zebra indexing -->
565 <xsl:output indent="yes" method="xml" version="1.0" encoding="UTF-8"/>
567 <!-- disable all default text node output -->
568 <xsl:template match="text()"/>
570 <!-- disable all default recursive element node transversal -->
571 <xsl:template match="node()"/>
573 <!-- match only on oai xml record root -->
574 <xsl:template match="/">
575 <z:record z:id="{normalize-space(oai:record/oai:header/oai:identifier)}">
576 <!-- you may use z:rank="{some XSLT; function here}" -->
578 <!-- explicetly calling defined templates -->
579 <xsl:apply-templates/>
583 <!-- OAI indexing templates -->
584 <xsl:template match="oai:record/oai:header/oai:identifier">
585 <z:index name="oai_identifier;0">
586 <xsl:value-of select="."/>
592 <!-- DC specific indexing templates -->
593 <xsl:template match="oai:record/oai:metadata/oai_dc:dc/dc:title">
594 <z:index name="dc_any:w dc_title:w dc_title:p dc_title:s ">
595 <xsl:value-of select="."/>
607 that the names and types of the indexes can be defined in the
608 indexing &xslt; stylesheet <emphasis>dynamically according to
609 content in the original &xml; records</emphasis>, which has
610 opportunities for great power and wizardry as well as grande
614 The following excerpt of a <emphasis>push</emphasis> stylesheet
615 <emphasis>might</emphasis>
616 be a good idea according to your strict control of the &xml;
617 input format (due to rigorous checking against well-defined and
618 tight RelaxNG or &xml; Schema's, for example):
621 <xsl:template name="element-name-indexes">
622 <z:index name="{name()}:w">
623 <xsl:value-of select="'1'"/>
628 This template creates indexes which have the name of the working
629 node of any input &xml; file, and assigns a '1' to the index.
631 <literal>find @attr 1=xyz 1</literal>
632 finds all files which contain at least one
633 <literal>xyz</literal> &xml; element. In case you can not control
634 which element names the input files contain, you might ask for
635 disaster and bad karma using this technique.
638 One variation over the theme <emphasis>dynamically created
639 indexes</emphasis> will definitely be unwise:
642 <!-- match on oai xml record root -->
643 <xsl:template match="/">
646 <!-- create dynamic index name from input content -->
647 <xsl:variable name="dynamic_content">
648 <xsl:value-of select="oai:record/oai:header/oai:identifier"/>
651 <!-- create zillions of indexes with unknown names -->
652 <z:index name="{$dynamic_content}:w">
653 <xsl:value-of select="oai:record/oai:metadata/oai_dc:dc"/>
660 Don't be tempted to play too smart tricks with the power of
661 &xslt;, the above example will create zillions of
662 indexes with unpredictable names, resulting in severe &zebra;
667 <section id="record-model-domxml-debug">
668 <title>Debuggig &dom; Filter Configurations</title>
670 It can be very hard to debug a &dom; filter setup due to the many
671 sucessive &marc; syntax translations, &xml; stream splitting and
672 &xslt; transformations involved. As an aid, you have always the
673 power of the <literal>-s</literal> command line switch to the
674 <literal>zebraidz</literal> indexing command at your hand:
676 zebraidx -s -c zebra.cfg update some_record_stream.xml
678 This command line simulates indexing and dumps a lot of debug
679 information in the logs, telling exactly which transformations
680 have been applied, how the documents look like after each
681 transformation, and which record ids and terms are send to the indexer.
686 <section id="record-model-domxml-elementset">
687 <title>&dom; Exchange Formats</title>
689 An exchange format can be anything which can be the outcome of an
690 &xslt; transformation, as far as the stylesheet is registered in
691 the main &dom; &xslt; filter configuration file, see
692 <xref linkend="record-model-domxml-filter"/>.
693 In principle anything that can be expressed in &xml;, HTML, and
694 TEXT can be the output of a <literal>schema</literal> or
695 <literal>element set</literal> directive during search, as long as
696 the information comes from the
697 <emphasis>original input record &xml; &dom; tree</emphasis>
698 (and not the transformed and <emphasis>indexed</emphasis> &xml;!!).
701 In addition, internal administrative information from the &zebra;
702 indexer can be accessed during record retrieval. The following
703 example is a summary of the possibilities:
706 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
707 xmlns:z="http://indexdata.dk/zebra/xslt/1"
710 <!- - register internal zebra parameters - ->
711 <xsl:param name="id" select="''"/>
712 <xsl:param name="filename" select="''"/>
713 <xsl:param name="score" select="''"/>
714 <xsl:param name="schema" select="''"/>
716 <xsl:output indent="yes" method="xml" version="1.0" encoding="UTF-8"/>
718 <!- - use then for display of internal information - ->
719 <xsl:template match="/">
721 <id><xsl:value-of select="$id"/></id>
722 <filename><xsl:value-of select="$filename"/></filename>
723 <score><xsl:value-of select="$score"/></score>
724 <schema><xsl:value-of select="$schema"/></schema>
737 <section id="record-model-domxml-example">
738 <title>&dom; Filter &oai; Indexing Example</title>
740 The source code tarball contains a working &dom; filter example in
741 the directory <filename>examples/dom-oai/</filename>, which
742 should get you started.
745 More example data can be harvested from any &oai; compliant server,
746 see details at the &oai;
747 <ulink url="http://www.openarchives.org/">
748 http://www.openarchives.org/</ulink> web site, and the community
750 <ulink url="http://www.openarchives.org/community/index.html">
751 http://www.openarchives.org/community/index.html</ulink>.
754 <ulink url="http://www.oaforum.org/tutorial/">
755 http://www.oaforum.org/tutorial/</ulink>.
767 <!-- Keep this comment at the end of the file
772 sgml-minimize-attributes:nil
773 sgml-always-quote-attributes:t
776 sgml-parent-document: "zebra.xml"
777 sgml-local-catalogs: nil
778 sgml-namecase-general:t