use schema and predicate types in extractors

1 files changed, 42 insertions, 11 deletions

diff --git a/test/extractor/generic/test_path.py b/test/extractor/generic/test_path.py
index 8623490..9376c7c 100644
--- a/test/extractor/generic/test_path.py
+++ b/test/extractor/generic/test_path.py
@@ -8,7 +8,9 @@ Author: Matthias Baumgartner, 2022
 import unittest
 
 # bsie imports
+from bsie import base
 from bsie.utils import ns
+from bsie.utils.bsfs import schema
 from bsie.utils.node import Node
 
 # objects to test
@@ -18,23 +20,52 @@ from bsie.extractor.generic.path import Path
 ## code ##
 
 class TestPath(unittest.TestCase):
+    def test_eq(self):
+        # distinct instances, same data
+        self.assertEqual(Path(), Path())
+        # different classes
+        class Foo(): pass
+        self.assertNotEqual(Path(), Foo())
+        self.assertNotEqual(Path(), 123)
+        self.assertNotEqual(Path(), None)
+
+    def test_schema(self):
+        self.assertEqual(Path().schema,
+            schema.Schema.from_string(base.extractor.SCHEMA_PREAMBLE + '''
+                bse:filename rdfs:subClassOf bsfs:Predicate ;
+                    rdfs:domain bsfs:Entity ;
+                    rdfs:range xsd:string ;
+                    owl:maxCardinality "INF"^^xsd:number .
+                '''))
+
     def test_extract(self):
-        node = Node(ns.bsfs.Entity, '') # Blank node
         ext = Path()
+        node = Node(ns.bsfs.Entity, '') # Blank node
+        content = '/tmp/foo/bar'
+        p_filename = ext.schema.predicate(ns.bse.filename)
+        entity = ext.schema.node(ns.bsfs.Node).get_child(ns.bsfs.Entity)
+        string = ext.schema.literal(ns.bsfs.Literal).get_child(ns.xsd.string)
 
         # baseline
-        self.assertSetEqual(set(ext.extract(node, '/tmp/foo/bar', (ns.bse.filename, ))),
-            {(node, ns.bse.filename, 'bar')})
+        self.assertSetEqual(set(ext.extract(node, content, (p_filename, ))),
+            {(node, p_filename, 'bar')})
         # predicates parameter is respected
-        self.assertSetEqual(set(ext.extract(node, '/tmp/foo/bar', (ns.bse.filename, ns.bse.foo))),
-            {(node, ns.bse.filename, 'bar')})
-        self.assertSetEqual(set(ext.extract(node, '/tmp/foo/bar', (ns.bse.foo, ))), set())
+        p_foo = ext.schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.foo, domain=entity, range=string) # unsupported predicate
+        self.assertSetEqual(set(ext.extract(node, content, (p_filename, p_foo))),
+            {(node, p_filename, 'bar')})
+        self.assertSetEqual(set(ext.extract(node, content, (p_foo, ))), set())
+        # predicates are validated
+        p_bar = p_foo.get_child(ns.bse.filename) # same URI but different hierarchy
+        self.assertSetEqual(set(ext.extract(node, content, (p_filename, p_bar))),
+            {(node, p_filename, 'bar')})
+        self.assertSetEqual(set(ext.extract(node, content, (p_bar, ))), set())
         # path variations
-        self.assertSetEqual(set(ext.extract(node, 'bar', (ns.bse.filename, ))),
-            {(node, ns.bse.filename, 'bar')})
-        self.assertSetEqual(set(ext.extract(node, '', (ns.bse.filename, ))),
-            {(node, ns.bse.filename, '')})
-        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.filename, ))), set())
+        self.assertSetEqual(set(ext.extract(node, 'bar', (p_filename, ))),
+            {(node, p_filename, 'bar')})
+        self.assertSetEqual(set(ext.extract(node, '', (p_filename, ))),
+            {(node, p_filename, '')})
+        # errors are suppressed
+        self.assertSetEqual(set(ext.extract(node, None, (p_filename, ))), set())
 
 
 ## main ##