Last updated: 2009-09-23.
This is a work-in-progress collection of test cases for RDFa in text/html. It has no official standing with any group.
The goal is that any RDFa-in-text/html implementation should extract the same set of triples from any given document (with no exceptions, even invalid documents and weird edge cases). This collection of tests tries to exercise a variety of 'interesting' cases, to help find bugs in implementations and in specifications.
There isn't yet a proper specification for RDFa-in-text/html. Each test has an "expected" result but it doesn't match any particular spec and shouldn't be taken too seriously. What is important is that any future RDFa-in-text/html specification must define precisely what the expected result is for each of these cases, and must also define whether each of these cases is valid or not.
See test results for several implementations. (Automatic colouring of results is only known to work in Firefox 3 and Opera 10. Disable stylesheets to see the raw results.)
See the test data (JSON).
(The "expected" value is basically N-Triples (but with <> for the current document),
and is the exact set of triples that must be produced.
The "id" value is an unstable identifier, so don't rely on it remaining unchanged.)
See also the live rdfQuery test runner and rdfa.js test runner. (Only known to work in Firefox 3 and Opera 10.)
The current tests roughly follow these groupings:
The test cases are copyright 2009 Philip Taylor, freely usable under the MIT license.