Slides

http://www.w3.org/2007/Talks/0808-extrememarkup-fs/

What is this talk about?

• Localization
• Schema Languages

What is Localization?

• your most favorite toys:
  • car
  • video game
  • markup schemes ...
• with needs for adaptation around the world!

What are Schema Languages?

• don't dare an deductive description for this audience
• let's be intensional:
  • XML DTDs
  • XML Schema
  • RELAX NG

What in Schema Languages can be localized?

• names (<p> to <段落>)
• data types　(2007-08-08 to 2007年8月8日)
• localization specific information (translate content of all <p> elements)

What is so special about these tasks?

Nothing! You can do everything with your favorite schema language

So why continue listening?

Three reasons follow:

(1) Using existing locale data

• Mainly the Common Locale Data Registry (CLDR)
• Provides locale data in Locale Data Markup Language (LDML) format
• Input to data type localization

(2) Applying the same information to various schema languages

• Specifying localization information once, using it multiple times
• Really a benefit? Depends on your point of view
  • Use existing mechansims of your most favorite schema language: no need for this paper
  • Combine and re-use necessary mechanisms: continue to listen

(3) Mapping of localized information to the locale unspecific representation

• Mapping of <段落> to <p>
• Mapping of 2007年8月8日 to 2007-08-08

Comparability of localized schemas

Now, how does it work?

Look into existing approaches to localization tasks. Then,

1. Identify a locale
2. Identify some markup (declarations)
3. Do the localization!

Wait! What about existing approaches?

Look into

• TEI ODD
  (renaming, documentation translation)
• CLDR
  (locale data)
• Internationalization Tag Set (ITS) 1.0
  (information for localization, e.g. translate attribute)

How to identify a locale?

First, what is a locale?

• set of user preferences
• depends on language, regions, scripts, timezones, etc.
• only rough definition possible :(
• see also Language and Locale Identifiers for the World Wide Web

But we need a locale identifier, right?

CLDR provides one, is based upon BCP 47 language identifiers

de_DE@collation=phonebook,currency=DDM

How does it work (again)?

1. Identify a locale
2. Identify some markup (declarations)
3. Do the localization!

Standoff usage

<loc:localInfo
 locale="de_DE"
 targetDeclaration="xs:element[@name='purchaseOrder']" ... >
 <loc:altIdent>Kaufbestellung</loc:altIdent>
</loc:localInfo>

Usage in a schema

<xs:element name="purchaseOrder">
    <xs:annotation>
        <xs:appinfo>
            <loc:localInfo locale="de_DE">
                <loc:altIdent>Kaufbestellung</loc:altIdent>
            </loc:localInfo>
        </xs:appinfo>
    </xs:annotation> ... </xs:element>

How does it work (again and again)?

1. Identify a locale
2. Identify some markup (declarations)
3. Do the localization!

We have seen element renaming

... and might also translate documentation

<xs:element name="purchaseOrder">
 <xs:annotation>
  <loc:localInfo locale="de_DE">
   <loc:altDocumentation>
 Schema zu Kaufbestellungen</loc:altDocumentation> ... </xs:element>

Modification of data types

Target date data type

<loc:localInfo locale="de">
 <loc:dateInfo calendarType="Gregorian" 
 dateFormatLengthType="long"/>
</loc:localInfo>

How to get data out of CLDR?

Using specific calendar types and date format types from CLDR

<dateFormatLength type="long">
 <dateFormat>
 <pattern>d. MMMM yyyy</pattern>
 </dateFormat>
</dateFormatLength>

What are these patterns in CLDR?

d. MMMM yyyy says:

• one or more digits for the day
• lexical item (full form) for the month
• four digits for the year
• Spaces inbetween

E.g., 8. August 2007

How can these CLDR patterns be used?

Input from CLDR mainly "d. MMMM yyyy"

Generating data type patterns for XML Schema data types

<xs:pattern
 value="\d{1,2}\.\s+(Januar|Februar|März|April|Mai|Juni|Juli|August
 |September|Oktober|November|Dezember)\s+\d{4}"/>

then, replace xs:date with the pattern

How does mapping to locale unspecific data work?

Input

• from CLDR mainly (again) "d. MMMM yyyy"
• locale specific date 8. August 2007

Output 2007-08-08

Show me some more CLDR data for data type modification

Locale display names for English en from CLDR

<localeDisplayNames>
  <languages>
   <language type="de">German</language> [...] </languages>
  <scripts>
   <script type="Latn">Latin</script> [...] </scripts>
  <territories>
   <territory type="DE">Germany</territory> [...] </territories>
 [...]</localeDisplayNames>

Show me some even more CLDR data!

Locale display names for German de from CLDR

 <localeDisplayNames>
  <languages>
   <language type="de">Deutsch</language> [...] </languages>
  <scripts>
   <script type="Latn">Lateinisch</script> [...] </scripts>
  <territories>
   <territory type="DE">Deutschland</territory> [...] </territories>
 [...]</localeDisplayNames>

How can such locale display names be used?

Generation of code lists

<xs:element name="language">
 <xs:annotation>
 <xs:appinfo>
  <loc:localInfo locale="de">
   <loc:localeDisplayNames type="languages"/>
  </loc:localInfo>
 </xs:appinfo>
</xs:annotation> ... </xs:element>

I might think it's great, if it really works ...

... it's being implemented with

• XSLT 2.0
• ANT script tasks
  • localize schema
  • de-localize instance
  • specific tasks for data type localization

Is it worth it?

• Use existing mechanisms of a specific schema language for specific localization tasks: no need for this paper
• Need for localization of various schemas and re-use of mechanisms: look into this

Questions?