Finish new RDF based ddf construction.
[htsworkflow.git] / htsworkflow / util / rdfhelp.py
1 """Helper features for working with librdf
2 """
3 import os
4 import types
5
6 import RDF
7
8 # standard ontology namespaces
9 owlNS = RDF.NS('http://www.w3.org/2002/07/owl#')
10 dublinCoreNS = RDF.NS("http://purl.org/dc/elements/1.1/")
11 rdfNS = RDF.NS("http://www.w3.org/1999/02/22-rdf-syntax-ns#")
12 rdfsNS= RDF.NS("http://www.w3.org/2000/01/rdf-schema#")
13 xsdNS = RDF.NS("http://www.w3.org/2001/XMLSchema#")
14
15 # internal ontologies
16 submissionOntology = RDF.NS("http://jumpgate.caltech.edu/wiki/UcscSubmissionOntology#")
17 dafTermOntology = RDF.NS("http://jumpgate.caltech.edu/wiki/UcscDaf#")
18 libraryOntology = RDF.NS("http://jumpgate.caltech.edu/wiki/LibraryOntology#")
19 submissionLog = RDF.NS("http://jumpgate.caltech.edu/wiki/SubmissionsLog/")
20
21 def blankOrUri(value=None):
22     node = None
23     if value is None:
24         node = RDF.Node()
25     elif type(value) in types.StringTypes:
26         node = RDF.Node(uri_string=value)
27     elif isinstance(value, RDF.Node):
28         node = value
29
30     return node
31
32
33 def toTypedNode(value):
34     if type(value) == types.BooleanType:
35         value_type = xsdNS['boolean'].uri
36         if value:
37             value = u'1'
38         else:
39             value = u'0'
40     elif type(value) in (types.IntType, types.LongType):
41         value_type = xsdNS['decimal'].uri
42         value = unicode(value)
43     elif type(value) == types.FloatType:
44         value_type = xsdNS['float'].uri
45         value = unicode(value)
46     elif type(value) in types.StringTypes:
47         value_type = xsdNS['string'].uri
48     else:
49         value_type = None
50         value = unicode(value)
51
52     return RDF.Node(literal=value, datatype=value_type)
53
54 def fromTypedNode(node):
55     if node is None:
56         return None
57
58     value_type = str(node.literal_value['datatype'])
59     # chop off xml schema declaration
60     value_type = value_type.replace(str(xsdNS[''].uri),'')
61     literal = node.literal_value['string']
62     literal_lower = literal.lower()
63
64     if value_type == 'boolean':
65         if literal_lower in ('1', 'yes', 'true'):
66             return True
67         elif literal_lower in ('0', 'no', 'false'):
68             return False
69         else:
70             raise ValueError("Unrecognized boolean %s" % (literal,))
71     elif value_type == 'decimal' and literal.find('.') == -1:
72         return int(literal)
73     elif value_type in ('decimal', 'float', 'double'):
74         return float(literal)
75     elif value_type in ('string'):
76         return literal
77     elif value_type in ('dateTime'):
78         raise NotImplemented('need to parse isoformat date-time')
79
80     return literal
81
82
83 def get_model(model_name=None, directory=None):
84     if directory is None:
85         directory = os.getcwd()
86         
87     if model_name is None:
88         storage = RDF.MemoryStorage()
89     else:
90         storage = RDF.HashStorage(model_name,
91                       options="hash-type='bdb',dir='{0}'".format(directory))
92     model = RDF.Model(storage)
93     return model
94         
95
96 def load_into_model(model, parser_name, filename, ns=None):
97     if not os.path.exists(filename):
98         raise IOError("Can't find {0}".format(filename))
99     
100     data = open(filename, 'r').read()
101     rdf_parser = RDF.Parser(name=parser_name)
102     rdf_parser.parse_string_into_model(model, data, ns)
103
104
105 def get_serializer(name='turtle'):
106     """Return a serializer with our standard prefixes loaded
107     """
108     writer = RDF.Serializer(name=name)
109     # really standard stuff
110     writer.set_namespace('owl', owlNS._prefix)
111     writer.set_namespace('rdf', rdfNS._prefix)
112     writer.set_namespace('rdfs', rdfsNS._prefix)
113     writer.set_namespace('xsd', xsdNS._prefix)
114
115     # should these be here, kind of specific to an application
116     writer.set_namespace('libraryOntology', libraryOntology._prefix)
117     writer.set_namespace('ucscSubmission', submissionOntology._prefix)
118     writer.set_namespace('ucscDaf', dafTermOntology._prefix)
119     return writer
120