6 from datetime import datetime
8 from htsworkflow.util.rdfhelp import \
16 guess_parser_by_extension, \
17 load_string_into_model, \
31 class TestRDFHelp(unittest.TestCase):
32 def test_from_none(self):
33 self.failUnlessEqual(fromTypedNode(None), None)
35 def test_typed_node_boolean(self):
36 node = toTypedNode(True)
37 self.failUnlessEqual(node.literal_value['string'], u'1')
38 self.failUnlessEqual(str(node.literal_value['datatype']),
39 'http://www.w3.org/2001/XMLSchema#boolean')
41 def test_bad_boolean(self):
42 node = RDF.Node(literal='bad', datatype=xsdNS['boolean'].uri)
43 self.failUnlessRaises(ValueError, fromTypedNode, node)
45 def test_typed_node_string(self):
46 node = toTypedNode('hello')
47 self.failUnlessEqual(node.literal_value['string'], u'hello')
48 self.failUnless(node.literal_value['datatype'] is None)
50 def test_typed_real_like(self):
52 node = toTypedNode(num)
53 self.failUnlessEqual(fromTypedNode(node), num)
55 def test_typed_integer(self):
57 node = toTypedNode(num)
58 self.failUnlessEqual(fromTypedNode(node), num)
59 self.failUnlessEqual(type(fromTypedNode(node)), type(num))
61 def test_typed_node_string(self):
64 self.failUnlessEqual(fromTypedNode(node), s)
65 self.failUnlessEqual(type(fromTypedNode(node)), types.UnicodeType)
67 def test_blank_or_uri_blank(self):
69 self.failUnlessEqual(node.is_blank(), True)
71 def test_blank_or_uri_url(self):
72 s = 'http://google.com'
74 self.failUnlessEqual(node.is_resource(), True)
75 self.failUnlessEqual(str(node.uri), s)
77 def test_blank_or_uri_node(self):
78 s = RDF.Node(RDF.Uri('http://google.com'))
80 self.failUnlessEqual(node.is_resource(), True)
81 self.failUnlessEqual(node, s)
83 def test_unicode_node_roundtrip(self):
85 roundtrip = fromTypedNode(toTypedNode(literal))
86 self.failUnlessEqual(roundtrip, literal)
87 self.failUnlessEqual(type(roundtrip), types.UnicodeType)
89 def test_datetime_no_microsecond(self):
90 dateTimeType = xsdNS['dateTime'].uri
91 short_isostamp = '2011-12-20T11:44:25'
92 short_node = RDF.Node(literal=short_isostamp,
93 datatype=dateTimeType)
94 short_datetime = datetime(2011,12,20,11,44,25)
96 self.assertEqual(fromTypedNode(short_node), short_datetime)
97 self.assertEqual(toTypedNode(short_datetime), short_node)
98 self.assertEqual(fromTypedNode(toTypedNode(short_datetime)),
101 def test_datetime_with_microsecond(self):
102 dateTimeType = xsdNS['dateTime'].uri
103 long_isostamp = '2011-12-20T11:44:25.081776'
104 long_node = RDF.Node(literal=long_isostamp,
105 datatype=dateTimeType)
106 long_datetime = datetime(2011,12,20,11,44,25,81776)
108 self.assertEqual(fromTypedNode(long_node), long_datetime)
109 self.assertEqual(toTypedNode(long_datetime), long_node)
110 self.assertEqual(fromTypedNode(toTypedNode(long_datetime)),
113 def test_strip_namespace_uri(self):
114 nsOrg = RDF.NS('example.org/example#')
115 nsCom = RDF.NS('example.com/example#')
119 self.failUnlessEqual(stripNamespace(nsOrg, node), term)
120 self.failUnlessEqual(stripNamespace(nsCom, node), None)
121 self.failUnlessEqual(stripNamespace(nsOrg, node.uri), term)
123 def test_strip_namespace_exceptions(self):
124 nsOrg = RDF.NS('example.org/example#')
125 nsCom = RDF.NS('example.com/example#')
127 node = toTypedNode('bad')
128 self.failUnlessRaises(ValueError, stripNamespace, nsOrg, node)
129 self.failUnlessRaises(ValueError, stripNamespace, nsOrg, nsOrg)
131 def test_simplify_uri(self):
132 DATA = [('http://asdf.org/foo/bar', 'bar'),
133 ('http://asdf.org/foo/bar#bleem', 'bleem'),
134 ('http://asdf.org/foo/bar/', 'bar'),
135 ('http://asdf.org/foo/bar?was=foo', 'was=foo')]
137 for uri, expected in DATA:
138 self.assertEqual(simplify_uri(uri), expected)
140 for uri, expected in DATA:
142 self.assertEqual(simplify_uri(n), expected)
144 for uri, expected in DATA:
145 n = RDF.Node(RDF.Uri(uri))
146 self.assertEqual(simplify_uri(n), expected)
148 # decoding literals is questionable
149 n = toTypedNode('http://foo/bar')
150 self.assertRaises(ValueError, simplify_uri, n)
152 def test_owl_import(self):
153 path, name = os.path.split(__file__)
154 #loc = 'file://'+os.path.abspath(path)+'/'
155 loc = os.path.abspath(path)+'/'
158 @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
159 @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
160 @prefix owl: <http://www.w3.org/2002/07/owl#> .
162 _:a owl:imports "{loc}extra.turtle" .
164 load_string_into_model(model, 'turtle', fragment, loc)
165 tc = RDF.Node(RDF.Uri('http://jumpgate.caltech.edu/wiki/TestCase'))
166 query = RDF.Statement(tc, rdfsNS['label'], None)
167 result = list(model.find_statements(query))
168 self.failUnlessEqual(len(result), 1)
169 self.failUnlessEqual(str(result[0].object), 'TestCase')
171 def test_sanitize_literal_text(self):
172 self.failUnlessRaises(ValueError, sanitize_literal, "hi")
174 hello_none = RDF.Node(hello_text)
175 self.failUnlessEqual(str(sanitize_literal(hello_none)),
177 hello_str = RDF.Node(literal=hello_text,
178 datatype=xsdNS['string'].uri)
179 hello_clean = sanitize_literal(hello_str)
180 self.failUnlessEqual(hello_clean.literal_value['string'],
183 def test_sanitize_literal_empty_string(self):
185 value_node = RDF.Node(value)
186 self.assertEqual(str(sanitize_literal(value_node)), value)
188 def test_sanitize_literal_html(self):
189 hello = "hello <a onload='javascript:alert(\"foo\");' href='http://google.com'>google.com</a>, whats up?"
190 hello_clean = 'hello <a href="http://google.com">google.com</a>, whats up?'
191 hello_node = RDF.Node(literal=hello,
192 datatype=xsdNS['string'].uri)
193 hello_sanitized = sanitize_literal(hello_node)
194 self.failUnlessEqual(hello_sanitized.literal_value['string'],
197 hostile = "hi <b>there</b><script type='text/javascript>alert('boo');</script><a href='javascript:alert('poke')>evil</a> scammer"
198 hostile_node = RDF.Node(hostile)
199 hostile_sanitized = sanitize_literal(hostile_node)
200 # so it drops the stuff after the javascript link.
201 # I suppose it could be worse
202 hostile_result = """hi <b>there</b>"""
203 self.failUnlessEqual(str(hostile_sanitized), hostile_result)
205 def test_guess_parser_from_file(self):
207 ('/a/b/c.rdf', 'rdfxml'),
208 ('/a/b/c.xml', 'rdfxml'),
209 ('/a/b/c.html', 'rdfa'),
210 ('/a/b/c.turtle', 'turtle'),
211 ('http://foo.bar/bleem.turtle', 'turtle')]
212 for path, parser in DATA:
213 self.assertEqual(guess_parser_by_extension(path), parser)
214 self.assertEqual(guess_parser(None, path), parser)
217 ('application/rdf+xml', 'http://a.org/b/c', 'rdfxml'),
218 ('application/x-turtle', 'http://a.org/b/c', 'turtle'),
219 ('text/html', 'http://a.org/b/c', 'rdfa'),
220 ('text/html', 'http://a.org/b/c.html', 'rdfa'),
221 ('text/plain', 'http://a.org/b/c.turtle', 'turtle'),
222 ('text/plain', 'http://a.org/b/c', 'guess')
224 for contenttype, url, parser in DATA:
225 self.assertEqual(guess_parser(contenttype, url), parser)
227 class TestRDFSchemas(unittest.TestCase):
228 def test_rdf_schema(self):
229 """Does it basically work?
232 self.assertEqual(model.size(), 0)
233 add_default_schemas(model)
234 self.assertGreater(model.size(), 0)
235 remove_schemas(model)
236 self.assertEqual(model.size(), 0)
238 def test_included_schemas(self):
240 add_default_schemas(model)
243 s = RDF.Statement(rdfNS[''], dcNS['title'], None)
244 title = model.get_target(rdfNS[''], dcNS['title'])
245 self.assertTrue(title is not None)
247 s = RDF.Statement(rdfNS['Property'], rdfNS['type'], rdfsNS['Class'])
248 self.assertTrue(model.contains_statement(s))
251 s = RDF.Statement(rdfsNS['Class'], rdfNS['type'], rdfsNS['Class'])
252 self.assertTrue(model.contains_statement(s))
254 s = RDF.Statement(owlNS['inverseOf'], rdfNS['type'],
256 self.assertTrue(model.contains_statement(s))
260 return unittest.makeSuite(TestRDFHelp, 'test')
261 except ImportError, e:
262 print "Unable to test rdfhelp"
267 if __name__ == "__main__":
268 unittest.main(defaultTest='suite')