@@ -414,6 +414,9 @@ def rdfnode(self, graph):
414414 selfnode = rdflib .URIRef (self .identity )
415415 for attr in self .attrs :
416416 objs = self .attrs [attr ]
417+ if (isinstance (objs , np .ndarray )):
418+ #try to convert np.ndarray to a list
419+ objs = objs .tolist ()
417420 if not (isinstance (objs , set ) or isinstance (objs , list )):
418421 objs = set ([objs ])
419422 for obj in objs :
@@ -438,6 +441,11 @@ def rdfgraph(self):
438441 """
439442 graph = rdflib .Graph ()
440443 graph .bind ('bald' , 'http://binary-array-ld.net/latest/' )
444+ for prefix_name in self ._prefixes :
445+ #strip the double underscore suffix
446+ new_name = prefix_name [:- 2 ]
447+
448+ graph .bind (new_name , self ._prefixes [prefix_name ])
441449 graph = self .rdfnode (graph )
442450
443451 return graph
@@ -530,75 +538,96 @@ def load(afilepath):
530538 finally :
531539 f .close ()
532540
533- def load_netcdf (afilepath , uri = None ):
541+ def load_netcdf (afilepath , baseuri = None ):
534542 """
535543 Validate a file with respect to binary-array-linked-data.
536544 Returns a :class:`bald.validation.Validation`
537545 """
538546
539547 with load (afilepath ) as fhandle :
540- prefix_group = (fhandle [fhandle .bald__isPrefixedBy ] if
548+ prefix_var_name = None
549+ if hasattr (fhandle , 'bald__isPrefixedBy' ):
550+ prefix_var_name = fhandle .bald__isPrefixedBy
551+
552+ prefix_var = (fhandle [fhandle .bald__isPrefixedBy ] if
541553 hasattr (fhandle , 'bald__isPrefixedBy' ) else {})
542554 prefixes = {}
543555
544556 skipped_variables = []
545- if prefix_group != {}:
546- prefixes = (dict ([(prefix , getattr (prefix_group , prefix )) for
547- prefix in prefix_group .ncattrs ()]))
548- if isinstance (prefix_group , netCDF4 ._netCDF4 .Variable ):
549- skipped_variables .append (prefix_group .name )
557+ if prefix_var != {}:
558+ prefixes = (dict ([(prefix , getattr (prefix_var , prefix )) for
559+ prefix in prefix_var .ncattrs ()]))
560+ if isinstance (prefix_var , netCDF4 ._netCDF4 .Variable ):
561+ skipped_variables .append (prefix_var .name )
550562 else :
551563 for k in fhandle .ncattrs ():
552564 if k .endswith ('__' ):
553565 prefixes [k ] = getattr (fhandle , k )
554- alias_group = (fhandle [fhandle .bald__isAliasedBy ]
566+
567+ # check that default set is handled, i.e. bald__ and rdf__
568+ if 'bald__' not in prefixes :
569+ prefixes ['bald__' ] = "http://binary-array-ld.net/latest/"
570+
571+ if 'rdf__' not in prefixes :
572+ prefixes ['rdf__' ] = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"
573+
574+ alias_var_name = None
575+ if hasattr (fhandle , 'bald__isAliasedBy' ):
576+ alias_var_name = fhandle .bald__isAliasedBy
577+
578+ alias_var = (fhandle [fhandle .bald__isAliasedBy ]
555579 if hasattr (fhandle , 'bald__isAliasedBy' ) else {})
556580 aliases = {}
557- if alias_group != {}:
558- aliases = (dict ([(alias , getattr (alias_group , alias ))
559- for alias in alias_group .ncattrs ()]))
560- if isinstance (alias_group , netCDF4 ._netCDF4 .Variable ):
561- skipped_variables .append (alias_group .name )
581+ if alias_var != {}:
582+ aliases = (dict ([(alias , getattr (alias_var , alias ))
583+ for alias in alias_var .ncattrs ()]))
584+ if isinstance (alias_var , netCDF4 ._netCDF4 .Variable ):
585+ skipped_variables .append (alias_var .name )
562586
563587 attrs = {}
564588 for k in fhandle .ncattrs ():
565589 attrs [k ] = getattr (fhandle , k )
566590 # It would be nice to use the URI of the file if it is known.
567- if uri is not None :
568- identity = uri
591+ if baseuri is not None :
592+ identity = baseuri
569593 else :
570594 identity = 'root'
571595 root_container = Container (identity , attrs , prefixes = prefixes ,
572596 aliases = aliases )
573597 root_container .attrs ['bald__contains' ] = []
574598 file_variables = {}
575599 for name in fhandle .variables :
600+ if name == prefix_var_name or name == alias_var_name :
601+ continue
576602
577603 sattrs = fhandle .variables [name ].__dict__ .copy ()
578604 # inconsistent use of '/'; fix it
579605 identity = name
606+ if baseuri is not None :
607+ identity = baseuri + "/" + name
580608
581609 # netCDF coordinate variable special case
582610 if (len (fhandle .variables [name ].dimensions ) == 1 and
583611 fhandle .variables [name ].dimensions [0 ] == name ):
584- sattrs ['bald__array' ] = name
612+ #sattrs['bald__array'] = name
613+ sattrs ['bald__array' ] = identity
585614 sattrs ['rdf__type' ] = 'bald__Reference'
615+
586616 if fhandle .variables [name ].shape :
587617 sattrs ['bald__shape' ] = fhandle .variables [name ].shape
588618 var = Array (identity , sattrs , prefixes = prefixes , aliases = aliases )
589619 else :
590620 var = Subject (identity , sattrs , prefixes = prefixes , aliases = aliases )
591- if name not in skipped_variables :
592- # Don't include skipped variables, such as prefix or alias
593- # variables, within the containment relation.
594- root_container .attrs ['bald__contains' ].append (var )
595-
621+ root_container .attrs ['bald__contains' ].append (var )
596622 file_variables [name ] = var
597623
598624
599625
600626 # cycle again and find references
601627 for name in fhandle .variables :
628+ if name == prefix_var_name or name == alias_var_name :
629+ continue
630+
602631 var = file_variables [name ]
603632 # reverse lookup based on type to be added
604633 lookups = ['bald__references' , 'bald__array' ]
@@ -624,6 +653,8 @@ def load_netcdf(afilepath, uri=None):
624653 # Else, define a bald:childBroadcast
625654 else :
626655 identity = '{}_{}_ref' .format (name , dim )
656+ if baseuri is not None :
657+ identity = baseuri + '/' + '{}_{}_ref' .format (name , dim )
627658 rattrs = {}
628659 rattrs ['rdf__type' ] = 'bald__Reference'
629660 reshape = [1 for adim in var_shape ]
0 commit comments