Issues in previous roundtrip that are fixed:
1. Roundtrip tests were skipping tests that would fail instead of
reporting them with xfail.
While this is maybe a minor distinction it
is more appropriate to report as xfail as that is a better semantic
description of a known issue. Skipping should be reserved for tests
that are expected to succeed when ran but can only run in a suitable
environment (e.g. if berkleydb is installed). Furthermore, marking
tests with xfail will still run the test, and will also report them
as unexpected pass (xpass) if they do actually pass.
2. Removed ("json-ld", "example-misc.n3") from xfail/skip as it does
pass.
Refactoring and improvements:
1. Removed `test/test_n3_suite.py::test_n3_writing`.
This test was doing something functionally equivalent to what
`test/test_roundtrip.py::test_n3` would do, except it was doing it
with different code (`test/testutils.py::check_serialize_parse`).
Specifically it would test, parse from n3, serialize to n3,
parse from n3, and this is also something that
`test/test_roundtrip.py::test_n3` does. One difference is that
`check_serialize_parse` was using `crapCompare` instead of
`rdflib.compare.isomorphic`, and `crapCompare` does a significantly
worse job but also hides an issue with `rdflib.compare.isomorphic`
and quoted graphs which is now correctly reported as xfail.
2. Removed `_get_test_files_formats` and `all_{n3,nt}_files` from
`test/test_{nt,n3}_suite.py` (and removed the files since there is
nothing else in them of value). These are now replaced with code
inside `test/test_roundtrip.py` which has less duplication and uses
`guess_format` for determinering formats instead of a custom if/elif
block.
3. Change skipping/xfail to work on file basename instead of path.
4. Changed file collector from a generator to a function returning a
list as the result of this function is iteratred over multiple times.
5. Changed test id's to be more meningful, e.g. `roundtrip_n3-writer-test-28.n3_n3_trig`
instead of `roundtrip-args303`.
Other changes:
1. Renamed `test/nt/test.ntriples` -> `test/nt/test.nt`: .ntriples is not
a recognized extension and the test this is being used in does not
test anything pertaining to extention handling.
2. Add guess_format tests for `.{rdf,nt,n3}` and negative tests for
`.{docx,mkv}` and a file with no extension.