This repository has been archived by the owner on Aug 30, 2018. It is now read-only.
-
Notifications
You must be signed in to change notification settings - Fork 12
/
tei-pm.odd
676 lines (629 loc) · 42 KB
/
tei-pm.odd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
<?xml version="1.0" encoding="utf-8"?>
<TEI xmlns="http://www.tei-c.org/ns/1.0" xmlns:skos="http://www.w3.org/2004/02/skos/core#" xml:lang="en">
<teiHeader>
<fileDesc>
<titleStmt>
<title>TEI ODD extensions for processing models</title>
</titleStmt>
<publicationStmt>
<publisher>TEI Consortium</publisher>
<availability>
<licence target="http://creativecommons.org/licenses/by-sa/3.0/"> Distributed under a Creative Commons
Attribution-ShareAlike 3.0 Unported License </licence>
<licence target="http://www.opensource.org/licenses/BSD-2-Clause">
<p>Copyright 2015 TEI Consortium.</p>
<p>All rights reserved.</p>
<p>Redistribution and use in source and binary forms, with or without modification, are permitted provided
that the following conditions are met:</p>
<list>
<item>Redistributions of source code must retain the above copyright notice, this list of conditions and
the following disclaimer.</item>
<item>Redistributions in binary form must reproduce the above copyright notice, this list of conditions
and the following disclaimer in the documentation and/or other materials provided with the
distribution.</item>
</list>
<p>This software is provided by the copyright holders and contributors "as is" and any express or implied
warranties, including, but not limited to, the implied warranties of merchantability and fitness for a
particular purpose are disclaimed. In no event shall the copyright holder or contributors be liable for
any direct, indirect, incidental, special, exemplary, or consequential damages (including, but not limited
to, procurement of substitute goods or services; loss of use, data, or profits; or business interruption)
however caused and on any theory of liability, whether in contract, strict liability, or tort (including
negligence or otherwise) arising in any way out of the use of this software, even if advised of the
possibility of such damage.</p>
</licence>
<p>TEI material can be licensed differently depending on the use you intend to make of it. Hence it is made
available under both the CC+BY and BSD-2 licences. The CC+BY licence is generally appropriate for usages
which treat TEI content as data or documentation. The BSD-2 licence is generally appropriate for usage of
TEI content in a software environment. For further information or clarification, please contact the <ref
target="mailto:info@tei-c.org">TEI Consortium</ref>. </p>
</availability>
</publicationStmt>
<sourceDesc>
<p>created ab initio during a meeting in Oxford; Significantly edited by James Cummings later.</p>
</sourceDesc>
</fileDesc>
<revisionDesc>
<change when="2015-09-11">Replaced elementspecs with xinclude links</change>
</revisionDesc>
</teiHeader>
<text>
<front>
<titlePage>
<docTitle>
<titlePart type="main">Proposed Guidelines Prose for TEI Processing Models</titlePart>
</docTitle>
<docAuthor>Sebastian Rahtz</docAuthor>
<docAuthor>James Cummings</docAuthor>
<docAuthor>Brian Pytlik Zillig</docAuthor>
<docAuthor>Magdalena Turska</docAuthor>
<docDate>Version 0.9: 29th May 2015</docDate>
</titlePage>
</front>
<body>
<div>
<head>Editing Note (to be removed)</head>
<p rend="italic bold">I (JamesC) believe that the appropriate place for this draft guidelines prose is 22.4.8
shortly after <ref target="http://www.tei-c.org/release/doc/tei-p5-doc/en/html/TD.html#TDENT"
>http://www.tei-c.org/release/doc/tei-p5-doc/en/html/TD.html#TDENT</ref> or another possibility is as a new
22.5 (pushing the current one down to 22.6).</p>
</div>
<div>
<head>Documenting a Processing Model</head>
<p>Where desired, the recording of the original rendition of source documents is catered for through use of the
global attributes (<att>rend</att>, <att>rendition</att> and <att>style</att>) as well as the
<gi>rendition</gi> element (see <ptr target="#COHQW"/>). What is needed in some other instances is a method
through which to document one or more possible 'processing models' for the document instance being encoded. In
this case a processing model is a template description for how the element should be transformed in one or
more particular outputs. It might say that this element should be displayed in italics or presented alongside
its alternative (e.g. in the example of <gi>sic</gi> and <gi>corr</gi>), with no details on how or in what
processing language this might be accomplished. It should be noted that any processing model documentation
should be considered illustrative rather than restrictive. This is a method for documenting one or more
<emph>possible</emph> processing models in abstract terms and it should be clear that processing of such
documents is not limited to only these possibilities. This enables the creator of a TEI customization to
specify how they intend particular elements could be processed. </p>
<p>This method of working <!-- informally known as the 'Rahtz Rationale' --> assumes a workflow with at least
three distinct parts and roles, in creating digital outputs from TEI texts. These roles may be held all by a
single individual or multiple large teams depending on the scope and available resources. The minimal
divisions that underlie this approach to documenting processing models are:</p>
<list type="ordered">
<item>an editor manages the text integrity and makes the high-level output decisions. These fall into (at
least) two parts:<list type="ordered">
<item>Structural descriptions, e.g. 'should the original or corrected version be displayed by default', or
'is this a block level or inline component'</item>
<item>Indications of appearance ('titles are in italics').</item>
</list></item>
<item>a programmer takes the editor's specification, and the TEI text(s), and creates the input for a designer
to make the output.</item>
<item>a designer creates the output envelope (for example, and book layout using typesetting software, or a
web site using their locally-supported web framework), making decisions in conjunction with the editor ('use
Garamond font throughout' or 'every page must show the departmental logo')</item>
</list>
<p>It is crucial to understand here that neither the designer nor the programmer necessarily need have any
understanding of the semantics of TEI elements or the structure of a TEI document. The programmer,
particularly, can use the processing model rules to extract material from the TEI text in a relatively simple
way. The method works for a variety of output formats, as the programmer simply has to provide implementations
for a small number of model behaviours, and understand the well-documented features of CSS. The concepts
addressed by the behaviours (e.g. <val>block</val> and <val>inline</val>) are generally in line with XSL-FO,
and HTML; the catalogue of functions is given under the element <gi>model</gi>.</p>
<p>The following elements are used to document processing models: <specList>
<specDesc key="model"/>
<specDesc key="modelSequence"/>
<specDesc key="modelGrp"/>
</specList>
</p>
<p>A processing model may by documented by one or more <gi>model</gi> elements. If these are provided within a
<gi>modelGrp</gi> element, or appear inside the <gi>elementSpec</gi> itself, then they document alternative
processing models. The <gi>modelGrp</gi> is used to group elements all of which are alternatives for the same
<att>output</att>. If multiple <gi>model</gi> elements are provided within a <gi>modelSequence</gi> element,
then they document a single model composed of a sequence of models to be processed as a unit.</p>
<div>
<head>Processing Models</head>
<p>The <gi>model</gi> element is the basis of any processing model documentation. The <gi>model</gi> element
is used to document the intended processing model for a particular element in an abstract manner which
separates the documentation of intended processing from its implementation in whatever processing language
is chosen. A required <att>behaviour</att> attribute which documents the intended processing behaviour and
controls which processing function should be used to present this element in output. By default the
behaviour specified by a <gi>model</gi> element is applicable to every occurrence of the element documented
by the parent <gi>elementSpec</gi> within which it appears. However, a <gi>model</gi> element's scope may be
modified using the <att>predicate</att> attribute to supply an XPath predicate determining the context under
which this model is applicable. </p>
<p>
<specList>
<specDesc key="model" atts="predicate behaviour useSourceRendition output"/>
<specDesc key="outputRendition" atts="scope"/>
</specList>
</p>
<p>The processing model notation allows each element to be assigned to one or more general categories
(<gi>model</gi>), and optionally given an output rendition description (<gi>outputRendition</gi>). This
allows a processor to know whether to handle the element or not, and broadly speaking how to display or
otherwise process it. The <gi>model</gi> and <gi>outputRendition</gi> documentation are recorded in the TEI
ODD customization. The editor or customization creator can document the intended output without having any
knowledge about the implementation language or methodologies employed. The scope of a <gi>model</gi> and its
<att>behaviour</att> may be limited by the <att>predicate</att> attribute. For a given
<gi>elementSpec</gi>, there can be as many <gi>model</gi> statements as required. Multiple <gi>model</gi>
statements are regarded as an alternation -- only one of them should be applied. This is comparable to a
'case statement' in some programming languages. </p>
<p>If the intention is for several <gi>model</gi> statements to be active at the same time, they can be
enclosed inside a <gi>modelSequence</gi>. If it is desired to group sets of <gi>model</gi> statements
without implying a sequence, for example grouping all of those for the same kind of output, then a
<gi>modelGrp</gi> element may be used.</p>
</div>
<div>
<head>Output Rendition </head>
<p>Within each <gi>model</gi> there can be set of <gi>outputRendition</gi> elements. For example, with the
simple case of handling of the <gi>hi</gi> element a customization writer may decide that this should be
treated as an inline element and normally output in italics:</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<model behaviour="inline">
<outputRendition>font-style: italic;</outputRendition>
</model>
</egXML>
<p>The <att>behaviour</att> attribute specifies the name of a model behaviour <val>inline</val>. This
behaviour and its parameters can be documented in a using the <gi>paramSpec</gi> element. While these
express different categories of output possibilities they do not specify the precise details of
implementation since that is dependent on the processing language chosen. The <gi>outputRendition</gi>
element uses CSS syntax to specify the appearance. It would have been possible here to simply add 'display:
inline;' to the rendition, to indicate that it is inline, but the intention is to separate the structural
decision from the presentation of its appearance. </p>
<p>It is common for <gi>hi</gi> to use a <att>rendition</att> attribute to indicate appearance. By default,
this is ignored, but <att>useSourceRendition</att> can be used to indicate that it should be taken account
of:</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<model behaviour="inline" useSourceRendition="true">
<outputRendition>font-style: italic;</outputRendition>
</model>
</egXML>
<p>In this case the <gi>outputRendition</gi> should be considered only a fallback and will be overridden by
the rendition attribute if it exists. This uses the same behaviour as CSS; i.e. if rendition involves a
change to a different property from 'font-style', then this local modification remains active.</p>
</div>
<div>
<head>Model Contexts</head>
<p>If the processing model is context-dependent, the <att>predicate</att> attribute can provide an XPath
expression to document when different models apply. In document processing for the <gi>q</gi> element, it
may be desired to indicate that when it is inside a paragraph it should be treated as <val>inline</val>
(with an italic font style), but when it isn't inside a paragraph, it should be treated as a block. </p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<model predicate="ancestor::p" behaviour="inline">
<outputRendition>font-style: italic;</outputRendition>
</model>
<model behaviour="block">
<outputRendition>left-margin: 2em;</outputRendition>
</model>
</egXML>
<p>The <gi>model</gi> elements here are alternatives and are so by default. It is only if the test (the
<att>predicate</att> attribute) does not match the context will the second <gi>model</gi> be applied. The
<gi>model</gi> elements could be grouped in a <gi>modelGrp</gi> element if there was a reason to do so,
for example they are all for one particular form of output. This enables a convenient distinction between
sets of models whose difference is not their predicate but their form of output.</p>
</div>
<div>
<head>Parameters for Model Behaviour</head>
<p>In the examples above TEI elements have been mapped to to simple block and inline concepts. There are other
behaviours possible and these can have parameters which themselves need to be documented. <specList>
<specDesc key="param" atts="name"/>
<specDesc key="paramList"/>
<specDesc key="paramSpec"/>
</specList>
</p>
<p>A more complex example is provided by processing the <gi>choice</gi> element; this is commonly used to
provide a pair of alternates (e.g. <gi>sic</gi> and <gi>corr</gi>). An editor or customization writer may
wish to document that the corrected version should be shown by default in the output, but still allow for
the implementation to switch to (or note or display in some form, such as a tooltip) the original version.
This may be documented with a <att>behaviour</att> attribute, for example, of <val>alternate</val> depending
on the function library being used. In documenting the processing model for <gi>choice</gi> there are three
common pairs of <gi>sic</gi> and <gi>corr</gi>, <gi>abbr</gi> and <gi>expan</gi>, as well as <gi>orig</gi>
and <gi>reg</gi> to record:</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<model predicate="sic and corr" behaviour="alternate">
<param name="default">corr</param>
<param name="alternate">sic</param>
</model>
<model predicate="abbr and expan[1]" behaviour="alternate">
<param name="default">expan</param>
<param name="alternate">abbr</param>
</model>
<model predicate="orig and reg" behaviour="alternate">
<param name="default">reg</param>
<param name="alternate">orig</param>
</model>
</egXML>
<p>This example provides explicit behaviour parameters as named <gi>param</gi> children of the <gi>model</gi>.
The <gi>param</gi> elements provide parameters to the behaviour implementation documented by the
<att>behaviour</att> attribute and their content is always assumed to be an XPath expression. Thus in the
example above, we specify <val>expan[1]</val> to choose the first of any available set of <gi>expan</gi>
children. The behaviours and their parameters may be documented separately for a customization using a
<gi>paramSpec</gi> element. If no <gi>param</gi> is supplied, or it is <val>.</val>, the content of the
object documented by this <gi>model</gi> and subject to the <att>behaviour</att> is assumed to be the
content of the current node including all child nodes of the element, its attribute nodes, text nodes, etc. </p>
<p>When a different behaviour is desired for other outputs this may be specified with the <att>output</att>
attribute. In the following example the processing intention for the <gi>date</gi> element is documented
where this has options when containing both regularized and human readable data. This example copes with
three situations but could be expanded. It records what to do when: </p>
<list type="ordered">
<item>there is text inside the element, and the output is print; </item>
<item>there is no text inside the element but there is a when attribute, and the output is print; and </item>
<item>there is a <att>when</att> attribute and text inside the element, and the output is web</item>
</list>
<p>In the third case, the same alternate function as in the previous example is used, but for print we
document that the <att>when</att> attribute should be used when there is no text.</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<model output="print" predicate="text()" behaviour="inline"/>
<model output="print" predicate="@when and not(text())" behaviour="inline">
<param name="content">@when</param>
</model>
<model predicate="@when" output="web" behaviour="alternate">
<param name="default">.</param>
<param name="alternate">@when</param>
</model>
</egXML>
<p>When documenting the output for elements with attributes, the processing model documentation can access
these attributes and their values using standard XPath notation (either abbreviated or unabbreviated
syntax). For example, the <gi>ref</gi> element usually has a <att>target</att> attribute providing the
location to which a link should be created. The <gi>ref</gi> element may or may not have text content, but
sometimes just the target. The processing model documentation below caters for these two situations:</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<model predicate="not(text())" behaviour="link">
<param name="content">@target</param>
<param name="link">@target</param>
</model>
<model behaviour="link">
<param name="content">.</param>
<param name="link">@target</param>
</model>
</egXML>
<p>Here the link function has two <gi>param</gi> elements, the text to be shown (<val>content</val>), and the
URL to which it should be linked (<val>link</val>). The first of these <gi>model</gi> elements documents
that when there is no text nodes inside the <gi>ref</gi> element it should use the content of the
<att>target</att> attribute <emph>as if</emph> it was textual content. This <att>predicate</att> attribute
could be made more specific as <val>not(text()) and @target</val> to assure that there is a
<att>target</att> attribute. The second of these uses this <att>target</att> attribute for providing the
link, but the element itself (and any text nodes or child elements) of the <gi>ref</gi> element for its
content.</p>
<p>The documentation for the suggested model behaviours is provided in the element specification for the
<gi>model</gi> element. For example, the <val>inline</val> behaviour has two optional parameters
<val>content</val> and <val>label</val>:</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<valItem ident="inline">
<paramList>
<paramSpec ident="content"/>
<paramSpec ident="label"/>
</paramList>
<desc versionDate="2015-08-21" xml:lang="en">creates inline element out
of <term>content</term> if there's something in <gi>outputRendition</gi>,
use that formatting otherwise just show text of selected content.</desc>
</valItem>
</egXML>
<p>This documentation inside the <gi>elementSpec</gi> for the <gi>model</gi> element because it is there that
we specify the list of attribute values that <val>behaviour</val> allows. Similarly the value for the
behaviour attribute <val>alternate</val> has two parameters: <val>alternate</val> and
<val>default</val>:</p>
<egXML xmlns="http://www.tei-c.org/ns/Examples">
<valItem ident="alternate">
<paramList>
<paramSpec ident="default">
<desc versionDate="2015-08-21" xml:lang="en">preferred content</desc>
</paramSpec>
<paramSpec ident="alternate">
<desc versionDate="2015-08-21" xml:lang="en">alternate content</desc>
</paramSpec>
</paramList>
<desc versionDate="2015-08-21" xml:lang="en">create a specialized display
of alternating elements for displaying the preferred version and an alternative,
both at once or by some method of toggling between the two.</desc>
</valItem>
</egXML>
<p>The nature of the documentation of these behaviour parameters intentionally makes no reference to the
processing functions which might implement since that is a decision that should be left to the implementor
and might vary by choice of processing language and form of output.</p>
</div>
<div>
<head>Implementing a TEI Processing Model</head>
<p>It is a principle of the processing model <!--(known informally as the 'Turska Tenet') --> that the
transformation stage should be minimally intelligent; that is to say, it should contain as few TEI-specific
rules as possible. The ODD should be as explicit as possible, and provide maximum expressivity to the
editor without recording details of implementation.</p>
<p>An implementation of a TEI Processing Model can take at least two approaches. One can read the TEI ODD
customization in advance and produce a processing pipeline (which itself consults the ODD or
not), or a processing implementation could consult the TEI ODD customization for information while processing a document
instance. Software which implements this processing model might proceed as follows:</p>
<list type="ordered">
<item>When processing an element in the instance document, any applicable <gi>model</gi> or
<gi>modelSequence</gi> for the current output format should be followed whether this is done by
consulting the TEI ODD customization directly or rules abstracted into the processing toolchain. </item>
<item>The <gi>model</gi> elements should be processed in order until a match with the predicate is found, or
a <gi>model</gi> with no predicate applies. If a <gi>model</gi> has no predicate, it matches any
occurrence of the element. If a <gi>model</gi> has no output, it applies to all forms of output. If there
is not a <att>useSourceRendition</att> attribute of <val>true</val> on the <gi>model</gi> then any
<att>rendition</att> attribute on the source element can be ignored.</item>
<item>All of the <gi>model</gi> elements inside a <gi>modelSequence</gi> should be applied in order if the
context matches any optional limiting predicates</item>
<item>The processing should follow the function provided by the <att>behaviour</att> attribute. Where the
parameter passed to that function is '.' or no <gi>param</gi> is supplied, then the object of the
<att>behaviour</att> attribute is all child nodes of the element including its attributes, element, and
text nodes. These nodes should be subject to processing within this context. Thus a block should create a
block-type object in the output for this element and all the results from processing its children.</item>
<item>If there are possible <gi>model</gi> elements, but none has a predicate which matches the current
element, processing should proceed to its child nodes and see if there are models for them. Likewise, if
there is no <gi>model</gi> provided for the current element, proceed to its children and see if there are
models for them.</item>
<item>The special behaviour <val>omit</val> should be interpreted to direct processing to stop for that
element and to proceed to the next sibling.</item>
</list>
<p>While this implementation system is lightly sketched it should be interpreted as broadly similar to the
processing model for XSLT.</p>
</div>
</div>
<!-- Previous text preserved in a comment here for sentimental and to enable my future checking on original intent -->
<!--<div>
<head>The TEI Processing Model</head>
<div>
<head>Introduction</head>
<p>TEI elements are generally
<emph>descriptive</emph> of the interpreted semantics of the
source text (<q>this is a title</q>, <q>this is a
quotation</q>), where the interpretation is often based on a
human interpretation of layout. The <att>rend</att>,
<att>rendition</att> and <att>style</att> attributes
are sometimes, but not necessarily, used to describe that
layout. The resulting encoded text is very amenable to
analysis, but for the common case of re-presenting the text on
the web or in print as a digital edition, the TEI model by itself is incomplete.</p>
<p>The processing model notation allows each element to be
assigned to a structural category (<gi>model</gi>), and given
an outline rendition description (<gi>outputRendition</gi>). This
allows a processor to know whether to handle the element or
not, and broadly speaking how to display or otherwise process
it. The model and rendition instructions are part of the
TEI ODD, so can be changed in TEI customizations in the normal
way.</p>
<p>
<figure>
<graphic url="SPM2.png"/>
<head>Processing model workflow</head>
</figure>
</p>
<p>This method of working (known informally as the Rahtz Rationale) assumes a workflow with three parts, and three
distinct roles, to create a digital edition from a TEI text:
<list rend="ordered"><item>an <emph>editor</emph> manages the text integrity,
makes the high-level output decisions. These fall into two parts
<list rend="ordered"><item><emph>Structural</emph> descriptions, eg <q>should the
original or corrected version be displayed by default</q>,
or <q>is this a block level or inline component</q></item><item>Indications of <emph>appearance</emph> (<q>titles are in
italics</q>).</item></list>
</item><item>a <emph>designer</emph> creates the output envelope (for
example, and book layout using InDesign, or a web site using
Drupal), making decisions in conjunction with the editor
(<q>use Garamond font throughout</q> or <q>every page must
show the departmental logo</q>)</item><item>a <emph>programmer</emph> takes the
<emph>editor</emph>'s specification, and the TEI text(s),
and creates the input for the <emph>designer</emph> to make
the output.</item></list>
It is crucial to understand here that neither the designer nor
the programmer need have any understanding of the semantics of
TEI elements, or the structure of a TEI document. The
programmer, particularly, can use the processing model rules
to extract material from the TEI text in a relatively simple
way. The method works for a variety of output formats, as the
programmer simply has to provide implementations for a small
number of functions, and understand the well-documented
features of CSS. The concepts addressed by the functions (eg
<q>block</q> and <q>inline</q>) are in line with XSL FO, and
HTML; the catalogue of functions is given under the element <gi>model</gi>.</p>
<p>A precondition for an
implementation is that it must include an XPath 2.0 interpreter,
as this is used in several places in the processing
notation. </p>
<p>It is a principle of the processing model (known informally as the
Turska Tenet) that the transformation stage should be
minimally intelligent; that is to say, it should contain as
few TEI-specific rules as possible. The ODD should be as
explicit as possible, and provide maximum expressivity to the
editor.</p>
</div>
<div>
<head>Processing model elements</head>
<p>There are two main elements for specifying the structural
and appearance components of a processing model:
<specList><specDesc key="model" atts="predicate behaviour useSourceRendition output"/><specDesc key="param" atts="name"/><specDesc key="outputRendition" atts="scope"/></specList>
For a given <gi>elementSpec</gi>, there can be as many
<gi>model</gi> statements as required. A set of multiple
<gi>model</gi> statements is regarded as an
<hi>alternation</hi>, i.e. only one of them will be
applied; this is comparable to a
<q>case statement</q> in some programming
languages. If the intention is for several
<gi>model</gi> statements to be active at the same time,
they can be enclosed inside a <gi>modelSequence</gi>.
</p>
<p>Within each <gi>model</gi> there can be set of
<gi>outputRendition</gi> elements.</p>
<p>Let us consider a simple case, the handling of the
<gi>hi</gi> element. We decide that it is an inline element,
normally rendered in italics:
<egXML xmlns="http://www.tei-c.org/ns/Examples"><model behaviour="inline"><outputRendition>font-style: italic;</outputRendition></model></egXML>
The <att>behaviour</att> attribute specifies the name of
a function, <term>inline</term>. If no <gi>param</gi>
is supplied to override it, the content of the inline
object is assumed to be the children of the element (text
or elements).
Separately, the <gi>outputRendition</gi> element uses CSS
syntax to specific the appearance. It would have been
possible here to simply add <code>display: inline</code>
to the rendition, to indicate that it is inline, but the
intention is to separate the structural decision from the
appearance decision.</p>
<p>It is common for <gi>hi</gi> to use a
<att>rendition</att> attribute to indicate appearance. By
default, this is ignored, but
<att>useSourceRendition</att> can be used to indicate
that it should be taken account of:
<egXML xmlns="http://www.tei-c.org/ns/Examples"><model behaviour="inline" useSourceRendition="true"><outputRendition>font-style: italic;</outputRendition></model></egXML>
in this case the <gi>outputRendition</gi> is only a fallback and
may be overridden by the <att>rendition</att> attribute. This uses
the same behaviour as CSS; ie if <att>rendition</att>
involves a change to a different property than
<q>font-style</q>, then this remains active.</p>
<p>If the processing model is context-dependent, we can
use <att>predicate</att> (an XPath 2.0 expression) to show
when different models apply:
<egXML xmlns="http://www.tei-c.org/ns/Examples"><model predicate="ancestor::p" behaviour="inline"><outputRendition>font-style: italic;</outputRendition></model><model behaviour="block"><outputRendition>left-margin: 2em;</outputRendition></model></egXML>
Here the <gi>model</gi>s are <emph>alternatives</emph>;
only if the test (the <att>predicate</att>) on the first
fails will the second be activated.</p>
<p>So far we have mapped TEI elements to simple block and
inline concepts. A more complex example is provided by
processing the <gi>choice</gi> element; this is commonly
used to provide a pair of alternates (eg <gi>sic</gi> and
<gi>corr</gi>), and our processing model has decided to
show
the corrected version by default, but allow for the
implementation to switch to, or display (perhaps as a
tooltip), the alternate; this service is provided by a
function <term>alternate</term>. We have three models, for
different cases of input:
<egXML xmlns="http://www.tei-c.org/ns/Examples"><model predicate="sic and corr" behaviour="alternate"><param name="default">corr</param><param name="alternate">sic</param></model><model predicate="abbr and expan[1]" behaviour="alternate"><param name="default">expan</param><param name="alternate">abbr</param></model><model predicate="orig and reg" behaviour="alternate"><param name="default">reg</param><param name="alternate">orig</param></model></egXML>
Here there are named <gi>param</gi> children of
<gi>model</gi>, which provide parameters to the function
given by <att>behaviour</att>. The content of the
parameters is always an XPath 2.0 expression. Thus in the
example above, we specify <code>expan[1]</code> to choose
the first of any available set of <gi>expan</gi>
children.</p>
<p>Sometimes we want different behaviour for different
output formats, specified using <att>output</att>; the
following example is designed to handle <gi>date</gi>,
which has an optional <att>when</att> containing a
regularized data. We allow for three situations: a) when there is text inside
the element, and we are making print; b) when there is no
text but there is a <att>when</att> attribute; and c) when
we are making web pages. In the latter case, we use the
same <term>alternate</term> function as in the previous example,
but for print we force use of <att>when</att> when it is present.
<egXML xmlns="http://www.tei-c.org/ns/Examples"><model output="print" predicate="text()" behaviour="inline"/><model output="print" predicate="@when and not(text())" behaviour="inline"><param name="content">@when</param></model><model predicate="@when" output="web" behaviour="alternate"><param name="default">.</param><param name="alternate">@when</param></model></egXML>
</p>
<p>Elements which have attributes can access the values
using the usual XPath 2.0 notation. For example, the
<gi>ref</gi> element has an important attribute,
<att>target</att>, to which is a link should be
created. Sometimes the <gi>ref</gi> has text content, but
sometimes just the <att>target</att>. Our rule caters for
these two situations:
<egXML xmlns="http://www.tei-c.org/ns/Examples"><model predicate="not(text())" behaviour="link"><param name="content">@target</param><param name="link">@target</param></model><model behaviour="link"><param name="content">.</param><param name="link">@target</param></model></egXML>
Here the <term>link</term> function has two <gi>param</gi>, the text
to be shown, and the URL to which it should be linked.</p>
</div>
</div>
<div>
<head>Implementing the TEI processing model</head>
<p>A program which implements this processing model should
proceed as follows:
<list rend="ordered"><item>When processing an element in the instance
document, any applicable <gi>model</gi> or
<gi>modelSequence</gi> for the current
output format should be followed. Process them in order until a match with
the <att>predicate</att> is found, or a <gi>model</gi>
with no <att>predicate</att> is available. If a
<gi>model</gi> has no <att>predicate</att>, it matches
any occurrence of the element. If a
<gi>model</gi> has no <att>output</att>, it applies to
all outputs.
</item><item>All of the <gi>model</gi>s inside a
<gi>modelSequence</gi> should be applied in order
(although they may have limiting <att>predicate</att>s</item><item>Follow the function provided by the
<att>behaviour</att> attribute. Where the value passed to
that function is <q>.</q>, follow this process for all
child elements of the current element, within the context
of the named function. Thus a <term>block</term>
should create a block-type object in the output
for this element and all the results from processing its children.</item><item>If there are possible <gi>model</gi>s, but none has a
<att>predicate</att> which matches the current element,
proceed to its children and see if there are models for
them.</item><item>If there is no <gi>model</gi> provided for the
element, proceed to its children and see if there are
models for them.</item><item>The special function <term>omit</term> means that
processing should stop for this element, and move to the
next sibling.</item></list>
</p>
<p>This is broadly similar to the processing model for XSLT. </p>
</div>
-->
</body>
<back>
<div>
<head>Extensions to the TEI</head>
<specGrp xml:id="additions">
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/param.xml"/>
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/paramSpec.xml"/>
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/paramList.xml"/>
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/model.xml"/>
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/modelSequence.xml"/>
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/modelGrp.xml"/>
<include xmlns="http://www.w3.org/2001/XInclude" href="processingModel/guidelines-specs/outputRendition.xml"/>
</specGrp>
</div>
<div>
<head>Changes to the TEI</head>
<specGrp xml:id="changes">
<!-- adding model/modelGrp to elementSpec element -->
<elementSpec ident="elementSpec" mode="change">
<content>
<sequence>
<alternate minOccurs="0" maxOccurs="unbounded">
<classRef key="model.glossLike"/>
<classRef key="model.descLike"/>
</alternate>
<classRef key="classes" minOccurs="0"/>
<classRef key="content" minOccurs="0"/>
<classRef key="valList" minOccurs="0"/>
<classRef key="constraintSpec" maxOccurs="unbounded" minOccurs="0"/>
<classRef key="attList" minOccurs="0"/>
<alternate maxOccurs="unbounded" minOccurs="0">
<elementRef key="model"/>
<elementRef key="modelGrp"/>
<elementRef key="modelSequence"/>
</alternate>
<classRef key="exemplum" maxOccurs="unbounded" minOccurs="0"/>
<classRef key="remarks" maxOccurs="unbounded" minOccurs="0"/>
<classRef key="listRef" maxOccurs="unbounded" minOccurs="0"/>
</sequence>
</content>
</elementSpec>
<elementSpec ident="valItem" mode="change" module="tagdocs">
<content>
<alternate maxOccurs="unbounded" minOccurs="0">
<classRef key="model.descLike"/>
<classRef key="model.glossLike"/>
</alternate>
<elementRef key="paramList" minOccurs="0" maxOccurs="1"/>
</content>
</elementSpec>
</specGrp>
</div>
<div>
<head>Schema for processing extended ODD</head>
<schemaSpec ident="teipm" start="TEI teiCorpus specGrp">
<moduleRef key="header"/>
<moduleRef key="core"/>
<moduleRef key="tei"/>
<moduleRef key="textstructure"/>
<moduleRef key="tagdocs"/>
<moduleRef key="figures"/>
<moduleRef key="linking"/>
<specGrpRef target="#additions"/>
<specGrpRef target="#changes"/>
<elementSpec ident="egXML" mode="change" module="tagdocs" ns="http://www.tei-c.org/ns/Examples">
<content>
<oneOrMore xmlns="http://relaxng.org/ns/structure/1.0">
<choice>
<text/>
<anyName/>
</choice>
</oneOrMore>
</content>
</elementSpec>
</schemaSpec>
</div>
</back>
</text>
</TEI>