1 from __future__ import print_function
5 from unittest import TestCase
6 from datetime import datetime
9 from htsworkflow.util.rdfhelp import \
10 add_default_schemas, \
17 guess_parser_by_extension, \
18 load_string_into_model, \
32 class TestRDFHelp(TestCase):
33 def test_from_none(self):
34 self.assertEqual(fromTypedNode(None), None)
36 def test_typed_node_boolean(self):
37 node = toTypedNode(True)
38 self.assertIn(node.literal_value['string'], (u'1', u'true'))
39 self.assertEqual(str(node.literal_value['datatype']),
40 'http://www.w3.org/2001/XMLSchema#boolean')
42 def test_bad_boolean(self):
43 node = RDF.Node(literal='bad', datatype=xsdNS['boolean'].uri)
44 # older versions of librdf ~< 1.0.16 left the literal
45 # alone. and thus should fail the fromTypedNode call
46 # newer versions coerced the odd value to false.
48 self.assertFalse(fromTypedNode(node))
49 except ValueError as e:
52 def test_typed_node_string_node_attributes(self):
53 node = toTypedNode('hello')
54 self.assertEqual(node.literal_value['string'], u'hello')
55 self.assertTrue(node.literal_value['datatype'] is None)
57 def test_typed_real_like(self):
59 node = toTypedNode(num)
60 self.assertEqual(fromTypedNode(node), num)
62 def test_typed_integer(self):
64 node = toTypedNode(num)
65 self.assertEqual(fromTypedNode(node), num)
66 self.assertEqual(type(fromTypedNode(node)), type(num))
68 def test_typed_node_string(self):
71 self.assertEqual(fromTypedNode(node), s)
72 self.assertTrue(isinstance(fromTypedNode(node), six.text_type))
74 def test_blank_or_uri_blank(self):
76 self.assertEqual(node.is_blank(), True)
78 def test_blank_or_uri_url(self):
79 s = 'http://google.com'
81 self.assertEqual(node.is_resource(), True)
82 self.assertEqual(str(node.uri), s)
84 def test_blank_or_uri_node(self):
85 s = RDF.Node(RDF.Uri('http://google.com'))
87 self.assertEqual(node.is_resource(), True)
88 self.assertEqual(node, s)
90 def test_unicode_node_roundtrip(self):
92 roundtrip = fromTypedNode(toTypedNode(literal))
93 self.assertTrue(isinstance(roundtrip, six.text_type))
95 def test_datetime_no_microsecond(self):
96 dateTimeType = xsdNS['dateTime'].uri
97 short_isostamp = '2011-12-20T11:44:25'
98 short_node = RDF.Node(literal=short_isostamp,
99 datatype=dateTimeType)
100 short_datetime = datetime(2011,12,20,11,44,25)
102 self.assertEqual(fromTypedNode(short_node), short_datetime)
103 self.assertEqual(toTypedNode(short_datetime), short_node)
104 self.assertEqual(fromTypedNode(toTypedNode(short_datetime)),
107 def test_datetime_with_microsecond(self):
108 dateTimeType = xsdNS['dateTime'].uri
109 long_isostamp = '2011-12-20T11:44:25.081776'
110 long_node = RDF.Node(literal=long_isostamp,
111 datatype=dateTimeType)
112 long_datetime = datetime(2011,12,20,11,44,25,81776)
114 self.assertEqual(fromTypedNode(long_node), long_datetime)
115 self.assertEqual(toTypedNode(long_datetime), long_node)
116 self.assertEqual(fromTypedNode(toTypedNode(long_datetime)),
119 def test_strip_namespace_uri(self):
120 nsOrg = RDF.NS('example.org/example#')
121 nsCom = RDF.NS('example.com/example#')
125 self.assertEqual(strip_namespace(nsOrg, node), term)
126 self.assertEqual(strip_namespace(nsCom, node), None)
127 self.assertEqual(strip_namespace(nsOrg, node.uri), term)
129 def test_strip_namespace_exceptions(self):
130 nsOrg = RDF.NS('example.org/example#')
131 nsCom = RDF.NS('example.com/example#')
133 node = toTypedNode('bad')
134 self.assertRaises(ValueError, strip_namespace, nsOrg, node)
135 self.assertRaises(ValueError, strip_namespace, nsOrg, nsOrg)
137 def test_simplify_uri(self):
138 DATA = [('http://asdf.org/foo/bar', 'bar'),
139 ('http://asdf.org/foo/bar#bleem', 'bleem'),
140 ('http://asdf.org/foo/bar/', 'bar'),
141 ('http://asdf.org/foo/bar?was=foo', 'was=foo')]
143 for uri, expected in DATA:
144 self.assertEqual(simplify_uri(uri), expected)
146 for uri, expected in DATA:
148 self.assertEqual(simplify_uri(n), expected)
150 for uri, expected in DATA:
151 n = RDF.Node(RDF.Uri(uri))
152 self.assertEqual(simplify_uri(n), expected)
154 # decoding literals is questionable
155 n = toTypedNode('http://foo/bar')
156 self.assertRaises(ValueError, simplify_uri, n)
158 def test_owl_import(self):
159 path, name = os.path.split(__file__)
160 #loc = 'file://'+os.path.abspath(path)+'/'
161 loc = os.path.abspath(path)+'/'
164 @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
165 @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
166 @prefix owl: <http://www.w3.org/2002/07/owl#> .
168 _:a owl:imports "{loc}extra.turtle" .
170 load_string_into_model(model, 'turtle', fragment, loc)
171 tc = RDF.Node(RDF.Uri('http://jumpgate.caltech.edu/wiki/TestCase'))
172 query = RDF.Statement(tc, rdfsNS['label'], None)
173 result = list(model.find_statements(query))
174 self.assertEqual(len(result), 1)
175 self.assertEqual(str(result[0].object), 'TestCase')
177 def test_sanitize_literal_text(self):
178 self.assertRaises(ValueError, sanitize_literal, "hi")
180 hello_none = RDF.Node(hello_text)
181 self.assertEqual(str(sanitize_literal(hello_none)),
183 hello_str = RDF.Node(literal=hello_text,
184 datatype=xsdNS['string'].uri)
185 hello_clean = sanitize_literal(hello_str)
186 self.assertEqual(hello_clean.literal_value['string'],
189 def test_sanitize_literal_empty_string(self):
191 value_node = RDF.Node(value)
192 self.assertEqual(str(sanitize_literal(value_node)), value)
194 def test_sanitize_literal_html(self):
195 hello = "hello <a onload='javascript:alert(\"foo\");' href='http://google.com'>google.com</a>, whats up?"
196 hello_clean = 'hello <a href="http://google.com">google.com</a>, whats up?'
197 hello_node = RDF.Node(literal=hello,
198 datatype=xsdNS['string'].uri)
199 hello_sanitized = sanitize_literal(hello_node)
200 self.assertEqual(hello_sanitized.literal_value['string'],
203 hostile = "hi <b>there</b><script type='text/javascript>alert('boo');</script><a href='javascript:alert('poke')>evil</a> scammer"
204 hostile_node = RDF.Node(hostile)
205 hostile_sanitized = sanitize_literal(hostile_node)
206 # so it drops the stuff after the javascript link.
207 # I suppose it could be worse
208 hostile_result = """hi <b>there</b>"""
209 self.assertEqual(str(hostile_sanitized), hostile_result)
211 def test_guess_parser_from_file(self):
213 ('/a/b/c.rdf', 'rdfxml'),
214 ('/a/b/c.xml', 'rdfxml'),
215 ('/a/b/c.html', 'rdfa'),
216 ('/a/b/c.turtle', 'turtle'),
217 ('http://foo.bar/bleem.turtle', 'turtle')]
218 for path, parser in DATA:
219 self.assertEqual(guess_parser_by_extension(path), parser)
220 self.assertEqual(guess_parser(None, path), parser)
223 ('application/rdf+xml', 'http://a.org/b/c', 'rdfxml'),
224 ('application/x-turtle', 'http://a.org/b/c', 'turtle'),
225 ('text/html', 'http://a.org/b/c', 'rdfa'),
226 ('text/html', 'http://a.org/b/c.html', 'rdfa'),
227 ('text/plain', 'http://a.org/b/c.turtle', 'turtle'),
228 ('text/plain', 'http://a.org/b/c', 'guess')
230 for contenttype, url, parser in DATA:
231 self.assertEqual(guess_parser(contenttype, url), parser)
233 class TestRDFSchemas(TestCase):
234 def test_rdf_schema(self):
235 """Does it basically work?
238 self.assertEqual(model.size(), 0)
239 add_default_schemas(model)
240 self.assertTrue(model.size() > 0)
241 remove_schemas(model)
242 self.assertEqual(model.size(), 0)
244 def test_included_schemas(self):
246 add_default_schemas(model)
249 s = RDF.Statement(rdfNS[''], dcNS['title'], None)
250 title = model.get_target(rdfNS[''], dcNS['title'])
251 self.assertTrue(title is not None)
253 s = RDF.Statement(rdfNS['Property'], rdfNS['type'], rdfsNS['Class'])
254 self.assertTrue(model.contains_statement(s))
257 s = RDF.Statement(rdfsNS['Class'], rdfNS['type'], rdfsNS['Class'])
258 self.assertTrue(model.contains_statement(s))
260 s = RDF.Statement(owlNS['inverseOf'], rdfNS['type'],
262 self.assertTrue(model.contains_statement(s))
265 except ImportError as e:
266 print("Unable to test rdfhelp")
269 from unittest import TestSuite, defaultTestLoader
271 suite.addTests(defaultTestLoader.loadTestsFromTestCase(TestRDFHelp))
272 suite.addTests(defaultTestLoader.loadTestsFromTestCase(TestRDFSchemas))
276 if __name__ == "__main__":
277 from unittest import main
278 main(defaultTest="suite")