Print

Print


Hi,

The idea of using quick fixes to guide end users into selecting the 
proper custom values for an attribute is quite interesting.

One more thing Oxygen 17.0 added was support to have a special XML 
configuration file which can add content completion values for elements 
or attributes (even by applying an XSLT stylesheet):

> http://www.oxygenxml.com/doc/versions/17.0/ug-editor//#topics/configuring-content-completion-proposals.html

Regards,
Radu

Radu Coravu
<oXygen/>  XML Editor, Schema Editor and XSLT Editor/Debugger
http://www.oxygenxml.com

On 6/25/2015 12:49 AM, [log in to unmask] wrote:
> Hi,
>
> On 24/06/2015 22:39, Martin Holmes wrote:
>>
>> I have a feeling this might now be possible in Oxygen -- check with
>> their support -- but up to now, we've been doing this in a sort of
>> roundabout way: I have ant build files which rebuild our ODD file
>> using all the values of personography @xml:ids as a valList for the
>> attribute concerned, and then build the schema for it. Then Oxygen
>> does prompt for the correct values. It's slightly annoying because it
>> means you have to run that job every time you modify the personography
>> before you can use a new id in the pointer attribute, but it does work.
>>
>
> I think I have found a slightly more flexible approach using Schematron
> Quick Fixes (https://github.com/schematron-quickfix/), which are
> supported since Oxygen 17 (see
> http://www.oxygenxml.com/demo/Schematron_Quick_Fixes.html).
>
> For example, this rudimentary Schematron file could do the job (untested!):
>
> <schema
>    xmlns="http://purl.oclc.org/dsdl/schematron"
>    xmlns:xsl="http://www.w3.org/1999/XSL/Transform"
>    xmlns:tei="http://www.tei-c.org/ns/1.0"
>    xmlns:sqf="http://www.schematron-quickfix.com/validator/process"
>    queryBinding="xslt2">
>      <ns prefix="tei" uri="http://www.tei-c.org/ns/1.0"/>
>    <xsl:variable name="person.index"
> select="doc('person-index.xml')//tei:person"/>
>    <pattern>
>      <rule context="tei:name">
>        <report role="warning" test="not(@ref)" sqf:fix="add.ref">
>          A name should have a @ref attribute pointing to a person
> definition.
>        </report>
>        <let name="person.id" value="$person.index[matches(tei:persName,
> normalize-space(current()))]/@xml:id"/>
>        <sqf:fix id="add.ref">
>          <sqf:description>
>            <sqf:title>
>              Add a @ref pointing to a person definition.
>            </sqf:title>
>          </sqf:description>
>          <sqf:add match="." node-type="attribute" target="ref"
> select="concat('#', $person.id)"/>
>        </sqf:fix>
>      </rule>
>    </pattern>
>
> </schema>
>
> The $person.index variable loads all person definitions from an index
> file; the Schematron rule matching <tei:name> elements will try to look
> up a person definition in $person.index matching the current name. Next,
> the SQF will propose to add a @ref attribute to the <tei:name> pointing
> to the the @xml:id value of the matching <tei:person> element in
> $person.index.
>
> Though this example is very rudimentary, you could add some degree of
> 'dynamism' to offer multiple suggestions, should multiple possible
> person definitions be found (see the discussion at
> https://github.com/schematron-quickfix/sqf/issues/3).
>
> While I think this is slightly more flexible than Martin's approach in
> that the ODD file doesn't have to be regenerated over again, the
> definition of $person.index in my example still requires a precompiled
> index file containing all unique names found in a TEI document or
> collection. While it is theoretically possible to do a live search over
> a collection (by swapping the doc() function with a collection() call
> over the entire collection), this would probably hardly be workable
> since each new Schematron validation run (basically after each keystroke
> in an XML document being validated against this Schematron in Oxygen)
> will have to scan the collection again for each name in that document.
> But perhaps Oxygen allows for database integration with e.g. an eXist-db
> instance that could boost performance of the $person.index lookup; I
> don't know if that's possible (if it is possible, I'd be interested to
> learn more).
>
> Best,
>
> Ron
>