Datatypes supported by sample implementation of TREX in Java

The sample implementation of TREX in Java supports XML Schema Part 2 as its datatyping language. The version of XML Schema currently supported is the Candidate Recommendation.

Named datatypes

Datatypes are supported as follows:

string: Fully supported.
boolean: Fully supported.
float: Fully supported.
double: Fully supported.
decimal: Fully supported.
timeDuration: Partially supported: validated like string.
recurringDuration: Not supported.
binary: Not supported.
uriReference: Partially supported: validated like string.
ID: Partially supported: validated like NCName.
IDREF: Partially supported: validated like NCName.
ENTITY: Partially supported: validated like NCName.
QName: Fully supported.
CDATA: Fully supported.
token: Fully supported.
language: Partially supported: validated like string.
IDREFS: Partially supported: validated like list of NCName.
ENTITIES: Partially supported: validated like list of NCName.
NMTOKEN: Fully supported.
NMTOKENS: Fully supported.
Name: Fully supported.
NCName: Fully supported.
NOTATION: Partially supported: validated like Name.
integer: Fully supported.
nonPositiveInteger: Fully supported.
negativeInteger: Fully supported.
long: Fully supported.
int: Fully supported.
short: Fully supported.
byte: Fully supported.
nonNegativeInteger: Fully supported.
unsignedLong: Fully supported.
unsignedInt: Fully supported.
unsignedShort: Fully supported.
unsignedByte: Fully supported.
positiveInteger: Fully supported.
timeInstant: Partially supported: validated like string.
time: Not supported.
timePeriod: Not supported.
date: Not supported.
month: Not supported.
year: Not supported.
century: Not supported.
recurringDate: Not supported.
recurringDay: Not supported.

Anonymous datatypes

Derivation of new datatypes by restriction, list and union is supported. The element bearing the trex:role="datatype" attribute must be a restriction, list, union or simpleType element.

Facets are supported as follows:

length: Fully supported.
minLength: Fully supported.
maxLength: Fully supported.
pattern: Fully supported, but requires external regular expression package.
enumeration: Fully supported.
whiteSpace: Fully supported.
maxInclusive: Fully supported.
maxExclusive: Fully supported.
minInclusive: Fully supported.
minExclusive: Fully supported.
precision: Fully supported.
scale: Fully supported.
encoding: Not supported.
duration: Not supported.
period: Not supported.

The implementation of the pattern facet requires an implementation of regular expressions as defined in XML Schema Part 2. The name of the class implementing the interface com.thaiopensource.datatype.xsd.RegexEngine must be specified as the value of the com.thaiopensource.trex.util.RegexEngine system property. If no such class is available, then any pattern will be assumed to match any string.

The class com.thaiopensource.trex.util.XercesRegexEngine implements com.thaiopensource.datatype.xsd.RegexEngine using the package org.apache.xerces.utils.regex which is included in the Xerces XML Parser. To use this, an option of

-Dcom.thaiopensource.trex.util.RegexEngine=com.thaiopensource.trex.util.XercesRegexEngine

would be added to the invocation of the java command. There is a trivial bug in the current (1.3.0) version of Xerces, which makes invalid regular expressions cause a NullPointerException. The bug report describes a workaround.

Alternatively, you can use IBM's Regex for Java package, available from Alphaworks. You can create a RegexEngine that uses this simply by changing org.apache.xerces.utils to com.ibm in com/thaiopensource/trex/util/XercesRegexEngine.java. The sample implementation does not include such a class because the license for Regex for Java is currently rather restrictive.