[Yanel-dev] customizing Yanel serialization formats: XHTML Strict Transitional, HTML Strict, (X)HTML 5, JSON, ...

[Guillaume Déflache]

Hi!

Michi asked me to document the ways to customize serialization 
serialization in Yanel, so here it is.

This can be done at two levels:
- in the RC files
- in Java code

First the following formats already exist in Yanel, so no need for 
customization to use them: XHTML Strict, HTML Transitional, XML, text.



To implement the following formats the following output properties must 
be set:


*HTML Strict*: set serializer key to HTML_TRANSITIONAL to get the base 
HTML properties right, then force the public doctype to the standard 
HTML Strict value: "-//W3C//DTD HTML 4.01//EN", the system doctype may 
be left alone as AFAIK mainstream browsers do not care and anyway Yanel 
still has a bug on that, see comment example below)

Example:
---8<---
<?xml version="1.0"?>

<yanel:resource-config xmlns:yanel="http://www.wyona.org/yanel/rti/1.0">
   [...]
   <yanel:custom-config>
     <views xmlns="http://www.wyona.org/yanel/rti/1.0">
       <view id="[...]">
         [...]
         <mime-type>text/html</mime-type>
         <serializer key="HTML_TRANSITIONAL">
         <!--XXX HACK: forcing the doctype manually since Yanel does not 
know about HTML Strict ATM: -->
           <doctype-public>-//W3C//DTD HTML 4.01//EN</doctype-public>
           <!-- does not seem to work (!?!), the system doctype is stuck 
with the 'loose' value: -->
 
<doctype-system>http://www.w3.org/TR/html4/strict.dtd</doctype-system>
<!--
           <indent>no</indent>
-->
         </serializer>
       </view>
       [...]
     </views>
   </yanel:custom-config>
</yanel:resource-config>
---8<---

*XHTML Transitional*: set serializer key to XHTML_STRICT to get the base 
XML properties right, then force the public doctype to the standard 
XHTML Transitional value: "-//W3C//DTD XHTML 1.0 Transitional//EN", the 
system doctype may remain wrong (same as above).

Sadly this alone does not quite work yet because of one Yanel bug and 
one missing workaround, see 
http://lists.wyona.org/pipermail/yanel-development/2010-June/004840.html

This can however been done in Java code by overriding 
#getViewDescriptor(String viewId) in every resource-type that needs the 
right format, using the following helper code example:
---8<---
     /**
      * Reconfigure the view to use XHTML 1.0 strict.
      */
     private static final void 
reconfigureViewDescriptor(ConfigurableViewDescriptor cvd) {
         cvd.setMimeType("text/html");
         cvd.setSerializerKey(SerializerFactory.XHTML_STRICT_KEY);
         Properties serializerProperties = new Properties();
         serializerProperties.setProperty(OutputKeys.INDENT, "no");
 
serializerProperties.setProperty(OutputKeys.OMIT_XML_DECLARATION, "yes");
         serializerProperties.setProperty(OutputKeys.ENCODING, "UTF-8");
         serializerProperties.setProperty(OutputKeys.DOCTYPE_PUBLIC, 
"-//W3C//DTD XHTML 1.0 Transitional//EN");
         serializerProperties.setProperty(OutputKeys.DOCTYPE_SYSTEM, 
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd");
         serializerProperties.setProperty(OutputKeys.METHOD, "html"); 
//XXX HACK: needed to force root element name in doctype declaration, 
see http://lists.wyona.org/pipermail/yanel-development/2010-June/004840.html
         cvd.setSerializerProperties(serializerProperties);
     }
---8<---


Other interesting format may be: JSON, HTML 5, XHTML 5, etc.


HTH,
    Guillaume