Shexer

Now you can parse gz/zip (local) source files without previously uncompressing them. sheXer won't write any content on your disk nor modify the used source files.

Gz files will be assumed to contain a single graph file (any RDF supported format). Zip files are assumed to contain one or more graphs files. In case of containing several files, they will be all parsed and merged for processing, but all the sources must have the same format (ttl, nt...)

You can now use all_classes mode when you are getting shapes from some SPARQL endpoint

There were error generating SHACL content out of RDF containing literals with some explicit type declarations. This has been fixed

Now sheXer can extract constraints with inverse_paths too. Example:

:Person
{
rdf:type [foaf:Person] ;
:likes IRI ? ;
}

:PopularPerson
{
rdf:type [foaf:Person] ;
^ :likes IRI ? ;
}

This means that a node conforming with the shape :PopularPerson must be foaf:Person type and also must be liked by at least someone. This is, for a node :p, there should be at least a triple such as (IRI, :like, :p) and another one such as (:p, a, foaf:Person) so :p conforms with :PopularPerson.

sheXer must be configured to extract inverse_paths when building the object Shaper. It's default behaviour is still extracting just direct constraints.

Yu can activate a verbose mode flag when calling shex_graph(). This will cause to get information about the shexing_process in via the standard output. This information is referred to the starting/ending time of different modules, and some metrics about the ammount of elements computed.

Also, some bugs has been corrected when parsing TTL files in iterative mode

The new parser is not fully compatible with the standard TTL yet. It works with some assumptions on frequent line jumps in turtle files.
If you choose TURTLE as input format, you'll still be using the Rdflib parser, ergo you won't be able to load big graphs into main memory. To use the new parser instead, chose TURTLE_ITER as input format.