implementing parameters to set graph uris and base uris

Author: jyucsiro

lib/bald/__init__.py

@@ -536,7 +536,7 @@ def load(afilepath):
     finally:
         f.close()
 
-def load_netcdf(afilepath, uri=None):
+def load_netcdf(afilepath, uri=None, baseuri=None):
     """
     Validate a file with respect to binary-array-linked-data.
     Returns a :class:`bald.validation.Validation`
@@ -600,11 +600,14 @@ def load_netcdf(afilepath, uri=None):
             sattrs = fhandle.variables[name].__dict__.copy()
             # inconsistent use of '/'; fix it
             identity = name
+            if baseuri is not None:
+                identity = baseuri + "/" + name
 
             # netCDF coordinate variable special case
             if (len(fhandle.variables[name].dimensions) == 1 and
                 fhandle.variables[name].dimensions[0] == name):
-                sattrs['bald__array'] = name
+                #sattrs['bald__array'] = name
+                sattrs['bald__array'] = identity
                 sattrs['rdf__type'] = 'bald__Reference'
 
             if fhandle.variables[name].shape:
@@ -648,6 +651,8 @@ def load_netcdf(afilepath, uri=None):
                         # Else, define a bald:childBroadcast
                         else:
                             identity = '{}_{}_ref'.format(name, dim)
+                            if baseuri is not None:
+                                identity = baseuri + '/' +  '{}_{}_ref'.format(name, dim)
                             rattrs = {}
                             rattrs['rdf__type'] = 'bald__Reference'
                             reshape = [1 for adim in var_shape]

nc2rdf/nc2rdf.py

@@ -6,21 +6,21 @@
 import numpy as np
 import bald
 
-def nc2rdf(ncfilename, outformat):  
+def nc2rdf(ncfilename, outformat, container_uri=None, default_baseuri=None):  
     #print("nc2rdf test")
     #print(ncfile)
-    root_container = bald.load_netcdf(ncfilename)
+    root_container = bald.load_netcdf(ncfilename, uri=container_uri, baseuri=default_baseuri)
     ttl = root_container.rdfgraph().serialize(format=outformat).decode("utf-8")
     print(ttl)
 
-def cdl2rdf(cdl_file, outformat): 
+def cdl2rdf(cdl_file, outformat, container_uri=None, default_baseuri=None): 
     #print("cdl2rdf test")
     #print(cdl_file)
     tfile, tfilename = tempfile.mkstemp('.nc')
     #print(tfilename)
     subprocess.check_call(['ncgen', '-o', tfilename, cdl_file])
 
-    nc2rdf(tfilename, outformat)
+    nc2rdf(tfilename, outformat, container_uri=container_uri, default_baseuri=default_baseuri)
 
     os.close(tfile)
     os.remove(tfilename)
@@ -29,15 +29,17 @@ def cdl2rdf(cdl_file, outformat):
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description='Convert netCDF metadata to RDF.')
     parser.add_argument('-o', action="store", dest="format", default='n3', help="RDF output format (n3 *default, ttl, xml)")
+    parser.add_argument('--uri', action="store", dest="uri", help="Container URI for the root of the graph")
+    parser.add_argument('--baseuri', action="store", dest="baseuri", help="Base URI for the graph")
     parser.add_argument('--cdl', action="store_true", dest="isCDL", default=False, help="Flag to indicate file is CDL")
     parser.add_argument('--nc', action="store_true", dest="isNC", default=False, help="Flag to indicate file is netCDF")
     parser.add_argument("ncfile", help="Path for the netCDF file")
 
     args = parser.parse_args()
 
     if(args.isCDL or args.ncfile.endswith(".cdl") or args.ncfile.endswith('.CDL')):
-        cdl2rdf(args.ncfile, args.format)
+        cdl2rdf(args.ncfile, args.format, container_uri=args.uri, default_baseuri=args.baseuri)
     elif(args.isNC or args.ncfile.endswith(".nc") or args.ncfile.endswith('.NC')):
-        nc2rdf(args.ncfile, args.format)
+        nc2rdf(args.ncfile, args.format, container_uri=args.uri, default_baseuri=args.baseuri)
     else:
         print("Unrecognised file suffix. Please indicate if CDL or NC via --cdl or --nc");