Release notes for pymetadata 0.6.4

pymetadata

We are pleased to release the next version of pymetadata including the
following changes:

Fixes

Omex.from_directory, and with it Omex.from_omex and Omex.from_url, add the entries in the order of the manifest, followed by the files which the manifest does not list, sorted by their location. The entries were added in the order of os.walk, which is the order of the file system, so manifest.entries and entries_by_format of the same archive differed between machines
RDFAnnotation.resource_normalized keeps the collection and the term of every resource: the result is a URL, and parsing it again gives the same collection and term. A collection which is not in the identifiers.org registry (http://identifiers.org/sabiork/1406, urn:miriam:foo:bar) was reduced to the bare term, and a term which does not match the pattern of its collection (http://identifiers.org/chebi/000000035) lost its collection. Both are now returned as https://identifiers.org/<collection>/<term>, and the compact identifier of a prefix which is not in the registry stays the URL https://identifiers.org/CMO:0000012 (#81)
identifiers.org URLs of a prefix with a hyphen or an underscore, like ec-code, pdb-ccd or go_ref, are parsed into collection and term instead of being kept as an arbitrary URL (#81)
the collection of a MIRIAM URN and of the collection/term notation is lowercased, as it already was for URLs, so TAXONOMY/9606 is the collection taxonomy (#81)

Your pymetadata team