Merge branch 'mb/tools' into develop

12 files changed, 759 insertions, 38 deletions

diff --git a/test/base/__init__.py b/test/base/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/base/__init__.py
diff --git a/test/base/test_extractor.py b/test/base/test_extractor.py
new file mode 100644
index 0000000..7a00079
--- /dev/null
+++ b/test/base/test_extractor.py
@@ -0,0 +1,70 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import unittest
+
+# bsie imports
+from bsie.utils import ns
+from bsie.utils.bsfs import schema as _schema, URI
+
+# objects to test
+from bsie.base import extractor
+
+
+## code ##
+
+class StubExtractor(extractor.Extractor):
+    def __init__(self):
+        super().__init__(_schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
+            bse:author rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+            bse:comment rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+        '''))
+
+    def extract(self, subject, content, predicates):
+        raise NotImplementedError()
+
+class StubSub(StubExtractor):
+    pass
+
+class TestExtractor(unittest.TestCase):
+    def test_essentials(self):
+        ext = StubExtractor()
+        self.assertEqual(str(ext), 'StubExtractor')
+        self.assertEqual(repr(ext), 'StubExtractor()')
+        self.assertEqual(ext, StubExtractor())
+        self.assertEqual(hash(ext), hash(StubExtractor()))
+
+        sub = StubSub()
+        self.assertEqual(str(sub), 'StubSub')
+        self.assertEqual(repr(sub), 'StubSub()')
+        self.assertEqual(sub, StubSub())
+        self.assertEqual(hash(sub), hash(StubSub()))
+        self.assertNotEqual(ext, sub)
+        self.assertNotEqual(hash(ext), hash(sub))
+
+    def test_predicates(self):
+        schema = _schema.Schema.Empty()
+        entity = schema.node(ns.bsfs.Node).get_child(ns.bsfs.Entity)
+        string = schema.literal(ns.bsfs.Literal).get_child(URI('http://www.w3.org/2001/XMLSchema#string'))
+        p_author = schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.author, domain=entity, range=string)
+        p_comment = schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.comment, domain=entity, range=string)
+        ext = StubExtractor()
+        self.assertSetEqual(set(ext.predicates()), {p_author, p_comment} | set(schema.predicates()))
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/base/test_reader.py b/test/base/test_reader.py
new file mode 100644
index 0000000..802b314
--- /dev/null
+++ b/test/base/test_reader.py
@@ -0,0 +1,45 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import unittest
+
+# objects to test
+from bsie.base import reader
+
+
+## code ##
+
+class StubReader(reader.Reader):
+    def __call__(self, path):
+        raise NotImplementedError()
+
+class StubSub(StubReader):
+    pass
+
+class TestReader(unittest.TestCase):
+    def test_essentials(self):
+        ext = StubReader()
+        self.assertEqual(str(ext), 'StubReader')
+        self.assertEqual(repr(ext), 'StubReader()')
+        self.assertEqual(ext, StubReader())
+        self.assertEqual(hash(ext), hash(StubReader()))
+
+        sub = StubSub()
+        self.assertEqual(str(sub), 'StubSub')
+        self.assertEqual(repr(sub), 'StubSub()')
+        self.assertEqual(sub, StubSub())
+        self.assertEqual(hash(sub), hash(StubSub()))
+        self.assertNotEqual(ext, sub)
+        self.assertNotEqual(hash(ext), hash(sub))
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/extractor/generic/test_constant.py b/test/extractor/generic/test_constant.py
index f3ab0a3..aa33fb4 100644
--- a/test/extractor/generic/test_constant.py
+++ b/test/extractor/generic/test_constant.py
@@ -20,39 +20,101 @@ from bsie.extractor.generic.constant import Constant
 class TestConstant(unittest.TestCase):
     def test_extract(self):
         schema = '''
-            bse:author a bsfs:Predicate ;
+            bse:author rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:Entity ;
                 rdfs:range xsd:string ;
                 owl:maxCardinality "1"^^xsd:number .
-
-            bse:comment a bsfs:Predicate ;
+            bse:comment rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:Entity ;
                 rdfs:range xsd:string ;
                 owl:maxCardinality "INF"^^xsd:number .
-
             '''
         tuples = [
             (ns.bse.author, 'Me, myself, and I'),
             (ns.bse.comment, 'the quick brown fox jumps over the lazy dog.'),
             ]
-        node = Node(ns.bsfs.Entity, '') # Blank node
-        predicates = (ns.bse.author, ns.bse.comment)
         ext = Constant(schema, tuples)
+        node = Node(ns.bsfs.Entity, '') # Blank node
+        p_author = ext.schema.predicate(ns.bse.author)
+        p_comment = ext.schema.predicate(ns.bse.comment)
+        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, None, predicates)),
-            {(node, pred, value) for pred, value in tuples})
+        self.assertSetEqual(set(ext.extract(node, None, (p_author, p_comment))),
+            {(node, p_author, 'Me, myself, and I'),
+             (node, p_comment, 'the quick brown fox jumps over the lazy dog.')})
         # predicates is respected
-        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.author, ns.bse.foobar))),
-            {(node, ns.bse.author, 'Me, myself, and I')})
-        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.comment, ns.bse.foobar))),
-            {(node, ns.bse.comment, 'the quick brown fox jumps over the lazy dog.')})
-        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.foobar, ns.bse.barfoo))), set())
-
-        # FIXME: should change!
-        # for now: no schema compliance
-        ext = Constant('', tuples)
-        self.assertSetEqual(set(ext.extract(node, None, predicates)),
-            {(node, pred, value) for pred, value in tuples})
+        p_foobar = ext.schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.foobar, domain=entity, range=entity)
+        self.assertSetEqual(set(ext.extract(node, None, (p_author, p_foobar))),
+            {(node, p_author, 'Me, myself, and I')})
+        self.assertSetEqual(set(ext.extract(node, None, (p_comment, p_foobar))),
+            {(node, p_comment, 'the quick brown fox jumps over the lazy dog.')})
+        p_barfoo = ext.schema.predicate(ns.bse.author).get_child(ns.bse.comment, domain=entity, range=string)
+        self.assertSetEqual(set(ext.extract(node, None, (p_foobar, p_barfoo))), set())
+
+    def test_construct(self):
+        # schema compliance
+        schema = '''
+            bse:author rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "1"^^xsd:number .
+            bse:comment rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+            '''
+        # can create a schema
+        self.assertIsInstance(Constant(schema, [
+            (ns.bse.author, 'Me, myself, and I'),
+            (ns.bse.comment, 'the quick brown fox jumps over the lazy dog.'),
+            ]), Constant)
+        # predicates are validated
+        self.assertRaises(KeyError, Constant, schema, [
+            (ns.bse.author, 'Me, myself, and I'),
+            (ns.bse.foobar, 'foobar!')])
+        # FIXME: values are validated
+        #class Foo(): pass # not string compatible
+        #self.assertRaises(ValueError, Constant, schema, [
+        #    (ns.bse.author, Foo())])
+
+    def test_eq(self):
+        schema_a = '''
+            bse:author rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "1"^^xsd:number .
+            '''
+        schema_b = '''
+            bse:comment rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "INF"^^xsd:number .
+            '''
+        tuples_a = [(ns.bse.author, 'Me, myself, and I')]
+        tuples_b = [(ns.bse.comment, 'the quick brown fox jumps over the lazy dog.') ]
+        # distinct instances, same data
+        self.assertEqual(
+            Constant(schema_a, tuples_a),
+            Constant(schema_a, tuples_a))
+        self.assertEqual(
+            hash(Constant(schema_a, tuples_a)),
+            hash(Constant(schema_a, tuples_a)))
+        # different data
+        self.assertNotEqual(
+            Constant(schema_a, tuples_a),
+            Constant(schema_b, tuples_b))
+        self.assertNotEqual(
+            hash(Constant(schema_a, tuples_a)),
+            hash(Constant(schema_b, tuples_b)))
+        # different objects
+        class Foo(): pass
+        self.assertNotEqual(Constant(schema_a, tuples_a), Foo())
+        self.assertNotEqual(hash(Constant(schema_a, tuples_a)), hash(Foo()))
+        self.assertNotEqual(Constant(schema_a, tuples_a), 123)
+        self.assertNotEqual(hash(Constant(schema_a, tuples_a)), hash(123))
+        self.assertNotEqual(Constant(schema_a, tuples_a), None)
+        self.assertNotEqual(hash(Constant(schema_a, tuples_a)), hash(None))
 
 
 ## main ##
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 ##
diff --git a/test/extractor/generic/test_stat.py b/test/extractor/generic/test_stat.py
index f89b053..26dad6a 100644
--- a/test/extractor/generic/test_stat.py
+++ b/test/extractor/generic/test_stat.py
@@ -9,7 +9,9 @@ import os
 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
@@ -18,21 +20,51 @@ from bsie.extractor.generic.stat import Stat
 
 ## code ##
 
-class TestConstant(unittest.TestCase):
+class TestStat(unittest.TestCase):
+    def test_eq(self):
+        # distinct instances, same data
+        self.assertEqual(Stat(), Stat())
+        # different classes
+        class Foo(): pass
+        self.assertNotEqual(Stat(), Foo())
+        self.assertNotEqual(Stat(), 123)
+        self.assertNotEqual(Stat(), None)
+
+    def test_schema(self):
+        self.assertEqual(Stat().schema,
+            schema.Schema.from_string(base.extractor.SCHEMA_PREAMBLE + '''
+                bse:filesize rdfs:subClassOf bsfs:Predicate ;
+                    rdfs:domain bsfs:Entity ;
+                    rdfs:range xsd:integer ;
+                    owl:maxCardinality "INF"^^xsd:number .
+                '''))
+
     def test_extract(self):
+        ext = Stat()
         node = Node(ns.bsfs.Entity, '') # Blank node
         content = os.stat(__file__)
-        ext = Stat()
+        p_filesize = ext.schema.predicate(ns.bse.filesize)
+        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, content, (ns.bse.filesize, ))),
-            {(node, ns.bse.filesize, content.st_size)})
+        self.assertSetEqual(set(ext.extract(node, content, (p_filesize, ))),
+            {(node, p_filesize, content.st_size)})
         # predicates parameter is respected
-        self.assertSetEqual(set(ext.extract(node, content, (ns.bse.filesize, ns.bse.foo))),
-            {(node, ns.bse.filesize, content.st_size)})
-        self.assertSetEqual(set(ext.extract(node, content, (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_filesize, p_foo))),
+            {(node, p_filesize, content.st_size)})
+        self.assertSetEqual(set(ext.extract(node, content, (p_foo, ))), set())
+        # predicates are validated
+        p_bar = p_foo.get_child(ns.bse.filesizse) # same URI but different hierarchy
+        self.assertSetEqual(set(ext.extract(node, content, (p_filesize, p_bar))),
+            {(node, p_filesize, content.st_size)})
+        self.assertSetEqual(set(ext.extract(node, content, (p_bar, ))), set())
         # content variations
-        self.assertSetEqual(set(ext.extract(node, None, (ns.bse.filesize, ))), set())
+        self.assertSetEqual(set(ext.extract(node, os.stat_result([12345] * len(content)), (p_filesize, p_bar))),
+            {(node, p_filesize, 12345)})
+        # errors are suppressed
+        self.assertSetEqual(set(ext.extract(node, None, (p_filesize, ))), set())
 
 
 ## main ##
diff --git a/test/tools/__init__.py b/test/tools/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/tools/__init__.py
diff --git a/test/tools/test_builder.py b/test/tools/test_builder.py
new file mode 100644
index 0000000..bef0e9d
--- /dev/null
+++ b/test/tools/test_builder.py
@@ -0,0 +1,247 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import logging
+import unittest
+
+# bsie imports
+from bsie import base
+from bsie.base import errors
+from bsie.utils.bsfs import URI
+
+# objects to test
+from bsie.tools.builder import ExtractorBuilder
+from bsie.tools.builder import PipelineBuilder
+from bsie.tools.builder import ReaderBuilder
+from bsie.tools.builder import _safe_load
+from bsie.tools.builder import _unpack_name
+
+
+## code ##
+
+class TestUtils(unittest.TestCase):
+    def test_safe_load(self):
+        # invalid module
+        self.assertRaises(errors.LoaderError, _safe_load, 'dBGHMSAYOoKeKMpywDoKZQycENFPvN', 'foobar')
+        self.assertRaises(errors.LoaderError, _safe_load, 'dBGHMSAYOoKeKMpywDoKZQycENFPvN.bar', 'foobar')
+        # partially valid module
+        self.assertRaises(errors.LoaderError, _safe_load, 'os.foo', 'foobar')
+        # invalid class
+        self.assertRaises(errors.LoaderError, _safe_load, 'os.path', 'foo')
+        # valid module and class
+        cls = _safe_load('collections.abc', 'Container')
+        import collections.abc
+        self.assertEqual(cls, collections.abc.Container)
+
+    def test_unpack_name(self):
+        self.assertRaises(TypeError, _unpack_name, 123)
+        self.assertRaises(TypeError, _unpack_name, None)
+        self.assertRaises(ValueError, _unpack_name, '')
+        self.assertRaises(ValueError, _unpack_name, 'path')
+        self.assertRaises(ValueError, _unpack_name, '.Path')
+        self.assertEqual(_unpack_name('path.Path'), ('path', 'Path'))
+        self.assertEqual(_unpack_name('path.foo.bar.Path'), ('path.foo.bar', 'Path'))
+
+
+class TestReaderBuilder(unittest.TestCase):
+    def test_build(self):
+        builder = ReaderBuilder({'bsie.reader.path.Path': {}})
+        # build configured reader
+        cls = builder.build('bsie.reader.path.Path')
+        import bsie.reader.path
+        self.assertIsInstance(cls, bsie.reader.path.Path)
+        # build unconfigured reader
+        cls = builder.build('bsie.reader.stat.Stat')
+        import bsie.reader.stat
+        self.assertIsInstance(cls, bsie.reader.stat.Stat)
+        # re-build previous reader (test cache)
+        self.assertEqual(cls, builder.build('bsie.reader.stat.Stat'))
+        # test invalid
+        self.assertRaises(TypeError, builder.build, 123)
+        self.assertRaises(TypeError, builder.build, None)
+        self.assertRaises(ValueError, builder.build, '')
+        self.assertRaises(ValueError, builder.build, 'Path')
+        self.assertRaises(errors.BuilderError, builder.build, 'path.Path')
+        # invalid config
+        builder = ReaderBuilder({'bsie.reader.stat.Stat': dict(foo=123)})
+        self.assertRaises(errors.BuilderError, builder.build, 'bsie.reader.stat.Stat')
+        builder = ReaderBuilder({'bsie.reader.stat.Stat': 123})
+        self.assertRaises(TypeError, builder.build, 'bsie.reader.stat.Stat')
+        # no instructions
+        builder = ReaderBuilder({})
+        cls = builder.build('bsie.reader.stat.Stat')
+        self.assertIsInstance(cls, bsie.reader.stat.Stat)
+
+
+
+class TestExtractorBuilder(unittest.TestCase):
+    def test_iter(self):
+        # no specifications
+        self.assertListEqual(list(ExtractorBuilder([])), [])
+        # some specifications
+        builder = ExtractorBuilder([
+            {'bsie.extractor.generic.path.Path': {}},
+            {'bsie.extractor.generic.stat.Stat': {}},
+            {'bsie.extractor.generic.path.Path': {}},
+            ])
+        self.assertListEqual(list(builder), [0, 1, 2])
+
+    def test_build(self):
+        # simple and repeated extractors
+        builder = ExtractorBuilder([
+            {'bsie.extractor.generic.path.Path': {}},
+            {'bsie.extractor.generic.stat.Stat': {}},
+            {'bsie.extractor.generic.path.Path': {}},
+            ])
+        ext = [builder.build(0), builder.build(1), builder.build(2)]
+        import bsie.extractor.generic.path
+        import bsie.extractor.generic.stat
+        self.assertListEqual(ext, [
+            bsie.extractor.generic.path.Path(),
+            bsie.extractor.generic.stat.Stat(),
+            bsie.extractor.generic.path.Path(),
+            ])
+        # out-of-bounds raises KeyError
+        self.assertRaises(IndexError, builder.build, 3)
+
+        # building with args
+        builder = ExtractorBuilder([
+            {'bsie.extractor.generic.constant.Constant': {
+                'schema': '''
+                    bse:author rdfs:subClassOf bsfs:Predicate ;
+                        rdfs:domain bsfs:Entity ;
+                        rdfs:range xsd:string ;
+                        owl:maxCardinality "1"^^xsd:number .
+                    bse:rating rdfs:subClassOf bsfs:Predicate ;
+                        rdfs:domain bsfs:Entity ;
+                        rdfs:range xsd:integer ;
+                        owl:maxCardinality "1"^^xsd:number .
+                    ''',
+                'tuples': [
+                    ('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I'),
+                    ('http://bsfs.ai/schema/Entity#rating', 123),
+                    ],
+            }}])
+        obj = builder.build(0)
+        import bsie.extractor.generic.constant
+        self.assertEqual(obj, bsie.extractor.generic.constant.Constant('''
+                    bse:author rdfs:subClassOf bsfs:Predicate ;
+                        rdfs:domain bsfs:Entity ;
+                        rdfs:range xsd:string ;
+                        owl:maxCardinality "1"^^xsd:number .
+                    bse:rating rdfs:subClassOf bsfs:Predicate ;
+                        rdfs:domain bsfs:Entity ;
+                        rdfs:range xsd:integer ;
+                        owl:maxCardinality "1"^^xsd:number .
+                    ''', [
+            ('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I'),
+            ('http://bsfs.ai/schema/Entity#rating', 123),
+            ]))
+
+        # building with invalid args
+        self.assertRaises(errors.BuilderError, ExtractorBuilder(
+            [{'bsie.extractor.generic.path.Path': {'foo': 123}}]).build, 0)
+        # non-dict build specification
+        self.assertRaises(TypeError, ExtractorBuilder(
+            [('bsie.extractor.generic.path.Path', {})]).build, 0)
+        # multiple keys per build specification
+        self.assertRaises(TypeError, ExtractorBuilder(
+            [{'bsie.extractor.generic.path.Path': {},
+              'bsie.extractor.generic.stat.Stat': {}}]).build, 0)
+        # non-dict value for kwargs
+        self.assertRaises(TypeError, ExtractorBuilder(
+            [{'bsie.extractor.generic.path.Path': 123}]).build, 0)
+
+
+
+
+class TestPipelineBuilder(unittest.TestCase):
+    def test_build(self):
+        prefix = URI('http://example.com/local/file#')
+        c_schema = '''
+            bse:author rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "1"^^xsd:number .
+            '''
+        c_tuples = [('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I')]
+        # prepare builders
+        rbuild = ReaderBuilder({})
+        ebuild = ExtractorBuilder([
+            {'bsie.extractor.generic.path.Path': {}},
+            {'bsie.extractor.generic.stat.Stat': {}},
+            {'bsie.extractor.generic.constant.Constant': dict(
+                schema=c_schema,
+                tuples=c_tuples,
+                )},
+            ])
+        # build pipeline
+        builder = PipelineBuilder(prefix, rbuild, ebuild)
+        pipeline = builder.build()
+        # delayed import
+        import bsie.reader.path
+        import bsie.reader.stat
+        import bsie.extractor.generic.path
+        import bsie.extractor.generic.stat
+        import bsie.extractor.generic.constant
+        # check pipeline
+        self.assertDictEqual(pipeline._ext2rdr, {
+            bsie.extractor.generic.path.Path(): bsie.reader.path.Path(),
+            bsie.extractor.generic.stat.Stat(): bsie.reader.stat.Stat(),
+            bsie.extractor.generic.constant.Constant(c_schema, c_tuples): None,
+            })
+
+        # fail to load extractor
+        ebuild_err = ExtractorBuilder([
+            {'bsie.extractor.generic.foo.Foo': {}},
+            {'bsie.extractor.generic.path.Path': {}},
+            ])
+        with self.assertLogs(logging.getLogger('bsie.tools.builder'), logging.ERROR):
+            pipeline = PipelineBuilder(prefix, rbuild, ebuild_err).build()
+            self.assertDictEqual(pipeline._ext2rdr, {
+                bsie.extractor.generic.path.Path(): bsie.reader.path.Path()})
+
+        # fail to build extractor
+        ebuild_err = ExtractorBuilder([
+            {'bsie.extractor.generic.path.Path': {'foo': 123}},
+            {'bsie.extractor.generic.path.Path': {}},
+            ])
+        with self.assertLogs(logging.getLogger('bsie.tools.builder'), logging.ERROR):
+            pipeline = PipelineBuilder(prefix, rbuild, ebuild_err).build()
+            self.assertDictEqual(pipeline._ext2rdr, {
+                bsie.extractor.generic.path.Path(): bsie.reader.path.Path()})
+
+        # fail to load reader
+        with self.assertLogs(logging.getLogger('bsie.tools.builder'), logging.ERROR):
+            # switch reader of an extractor
+            old_reader = bsie.extractor.generic.path.Path.CONTENT_READER
+            bsie.extractor.generic.path.Path.CONTENT_READER = 'bsie.reader.foo.Foo'
+            # build pipeline with invalid reader reference
+            pipeline = PipelineBuilder(prefix, rbuild, ebuild).build()
+            self.assertDictEqual(pipeline._ext2rdr, {
+                bsie.extractor.generic.stat.Stat(): bsie.reader.stat.Stat(),
+                bsie.extractor.generic.constant.Constant(c_schema, c_tuples): None,
+                })
+            # switch back
+            bsie.extractor.generic.path.Path.CONTENT_READER = old_reader
+
+        # fail to build reader
+        rbuild_err = ReaderBuilder({'bsie.reader.stat.Stat': dict(foo=123)})
+        with self.assertLogs(logging.getLogger('bsie.tools.builder'), logging.ERROR):
+            pipeline = PipelineBuilder(prefix, rbuild_err, ebuild).build()
+            self.assertDictEqual(pipeline._ext2rdr, {
+                bsie.extractor.generic.path.Path(): bsie.reader.path.Path(),
+                bsie.extractor.generic.constant.Constant(c_schema, c_tuples): None,
+                })
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/tools/test_pipeline.py b/test/tools/test_pipeline.py
new file mode 100644
index 0000000..9888d2e
--- /dev/null
+++ b/test/tools/test_pipeline.py
@@ -0,0 +1,167 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import logging
+import os
+import unittest
+
+# bsie imports
+from bsie.base import errors
+from bsie.utils import ns
+from bsie.utils.bsfs import URI
+from bsie.utils.node import Node
+import bsie.extractor.generic.constant
+import bsie.extractor.generic.path
+import bsie.extractor.generic.stat
+import bsie.reader.path
+import bsie.reader.stat
+
+# objects to test
+from bsie.tools.pipeline import Pipeline
+
+
+## code ##
+
+class TestPipeline(unittest.TestCase):
+    def setUp(self):
+        # constant A
+        csA = '''
+            bse:author rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                owl:maxCardinality "1"^^xsd:number .
+            '''
+        tupA = [('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I')]
+        # constant B
+        csB = '''
+            bse:rating rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:integer ;
+                owl:maxCardinality "1"^^xsd:number .
+            '''
+        tupB = [('http://bsfs.ai/schema/Entity#rating', 123)]
+        # extractors/readers
+        self.ext2rdr = {
+            bsie.extractor.generic.path.Path(): bsie.reader.path.Path(),
+            bsie.extractor.generic.stat.Stat(): bsie.reader.stat.Stat(),
+            bsie.extractor.generic.constant.Constant(csA, tupA): None,
+            bsie.extractor.generic.constant.Constant(csB, tupB): None,
+        }
+        self.prefix = URI('http://example.com/local/file#')
+
+    def test_essentials(self):
+        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        self.assertEqual(str(pipeline), 'Pipeline')
+        self.assertEqual(repr(pipeline), 'Pipeline(...)')
+
+    def test_equality(self):
+        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        # a pipeline is equivalent to itself
+        self.assertEqual(pipeline, pipeline)
+        self.assertEqual(hash(pipeline), hash(pipeline))
+        # identical builds are equivalent
+        self.assertEqual(pipeline, Pipeline(self.prefix, self.ext2rdr))
+        self.assertEqual(hash(pipeline), hash(Pipeline(self.prefix, self.ext2rdr)))
+
+        # equivalence respects prefix
+        self.assertNotEqual(pipeline, Pipeline(URI('http://example.com/global/ent#'), self.ext2rdr))
+        self.assertNotEqual(hash(pipeline), hash(Pipeline(URI('http://example.com/global/ent#'), self.ext2rdr)))
+        # equivalence respects extractors/readers
+        ext2rdr = {ext: rdr for idx, (ext, rdr) in enumerate(self.ext2rdr.items()) if idx % 2 == 0}
+        self.assertNotEqual(pipeline, Pipeline(self.prefix, ext2rdr))
+        self.assertNotEqual(hash(pipeline), hash(Pipeline(self.prefix, ext2rdr)))
+
+        # equivalence respects schema
+        p2 = Pipeline(self.prefix, self.ext2rdr)
+        p2.schema = pipeline.schema.Empty()
+        self.assertNotEqual(pipeline, p2)
+        self.assertNotEqual(hash(pipeline), hash(p2))
+
+        # not equal to other types
+        class Foo(): pass
+        self.assertNotEqual(pipeline, Foo())
+        self.assertNotEqual(hash(pipeline), hash(Foo()))
+        self.assertNotEqual(pipeline, 123)
+        self.assertNotEqual(hash(pipeline), hash(123))
+        self.assertNotEqual(pipeline, None)
+        self.assertNotEqual(hash(pipeline), hash(None))
+
+
+    def test_call(self):
+        # build pipeline
+        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        # build objects for tests
+        content_hash = 'e3bb4ab54e4a50d75626a1f76814f152f4edc60a82ad724aa2aa922ca5534427'
+        subject = Node(ns.bsfs.Entity, self.prefix + content_hash)
+        testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
+        p_filename = pipeline.schema.predicate(ns.bse.filename)
+        p_filesize = pipeline.schema.predicate(ns.bse.filesize)
+        p_author = pipeline.schema.predicate(ns.bse.author)
+        p_rating = pipeline.schema.predicate(ns.bse.rating)
+        entity = pipeline.schema.node(ns.bsfs.Entity)
+        p_invalid = pipeline.schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.foo, range=entity)
+
+        # extract given predicates
+        self.assertSetEqual(set(pipeline(testfile, {p_filename, p_filesize})), {
+            (subject, p_filename, 'testfile.t'),
+            (subject, p_filesize, 11),
+            })
+        self.assertSetEqual(set(pipeline(testfile, {p_author})), {
+            (subject, p_author, 'Me, myself, and I'),
+            })
+        self.assertSetEqual(set(pipeline(testfile, {p_filename})), {
+            (subject, p_filename, 'testfile.t'),
+            })
+        self.assertSetEqual(set(pipeline(testfile, {p_filesize})), {
+            (subject, p_filesize, 11),
+            })
+        # extract all predicates
+        self.assertSetEqual(set(pipeline(testfile)), {
+            (subject, p_filename, 'testfile.t'),
+            (subject, p_filesize, 11),
+            (subject, p_author, 'Me, myself, and I'),
+            (subject, p_rating, 123),
+            })
+        # invalid predicate
+        self.assertSetEqual(set(pipeline(testfile, {p_invalid})), set())
+        # valid/invalid predicates mixed
+        self.assertSetEqual(set(pipeline(testfile, {p_filename, p_invalid})), {
+            (subject, p_filename, 'testfile.t'),
+            })
+        # invalid path
+        self.assertRaises(FileNotFoundError, list, pipeline('inexistent_file'))
+        # FIXME: unreadable file (e.g. permissions error)
+
+    def test_call_reader_err(self):
+        class FaultyReader(bsie.reader.path.Path):
+            def __call__(self, path):
+                raise errors.ReaderError('reader error')
+
+        pipeline = Pipeline(self.prefix, {bsie.extractor.generic.path.Path(): FaultyReader()})
+        with self.assertLogs(logging.getLogger('bsie.tools.pipeline'), logging.ERROR):
+            testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
+            p_filename = pipeline.schema.predicate(ns.bse.filename)
+            self.assertSetEqual(set(pipeline(testfile, {p_filename})), set())
+
+    def test_call_extractor_err(self):
+        class FaultyExtractor(bsie.extractor.generic.path.Path):
+            def extract(self, subject, content, predicates):
+                raise errors.ExtractorError('extractor error')
+
+        pipeline = Pipeline(self.prefix, {FaultyExtractor(): bsie.reader.path.Path()})
+        with self.assertLogs(logging.getLogger('bsie.tools.pipeline'), logging.ERROR):
+            testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
+            p_filename = pipeline.schema.predicate(ns.bse.filename)
+            self.assertSetEqual(set(pipeline(testfile, {p_filename})), set())
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/tools/testfile.t b/test/tools/testfile.t
new file mode 100644
index 0000000..58bf1b8
--- /dev/null
+++ b/test/tools/testfile.t
@@ -0,0 +1 @@
+hello worl
diff --git a/test/utils/__init__.py b/test/utils/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/utils/__init__.py
diff --git a/test/utils/test_node.py b/test/utils/test_node.py
new file mode 100644
index 0000000..826f199
--- /dev/null
+++ b/test/utils/test_node.py
@@ -0,0 +1,66 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# imports
+import unittest
+
+# bsie imports
+from bsie.utils.bsfs import URI
+from bsie.utils import ns
+
+# objects to test
+from bsie.utils.node import Node
+
+
+## code ##
+
+class TestNode(unittest.TestCase):
+    def test_equality(self):
+        uri = URI('http://example.com/me/entity#1234')
+        node = Node(ns.bsfs.Entity, uri)
+        # basic equivalence
+        self.assertEqual(node, Node(ns.bsfs.Entity, URI('http://example.com/me/entity#1234')))
+        self.assertEqual(hash(node), hash(Node(ns.bsfs.Entity, URI('http://example.com/me/entity#1234'))))
+        # equality respects uri
+        self.assertNotEqual(node, Node(ns.bsfs.Entity, URI('http://example.com/me/entity#4321')))
+        self.assertNotEqual(hash(node), hash(Node(ns.bsfs.Entity, URI('http://example.com/me/entity#4321'))))
+        # equality respects node_type
+        self.assertNotEqual(node, Node(ns.bsfs.Foo, uri))
+        self.assertNotEqual(hash(node), hash(Node(ns.bsfs.Foo, uri)))
+        # not equal to other types
+        self.assertNotEqual(node, 1234)
+        self.assertNotEqual(hash(node), hash(1234))
+        self.assertNotEqual(node, uri)
+        self.assertNotEqual(hash(node), hash(uri))
+        self.assertNotEqual(node, ns.bsfs.Entity)
+        self.assertNotEqual(hash(node), hash(ns.bsfs.Entity))
+        class Foo(): pass
+        self.assertNotEqual(node, Foo())
+        self.assertNotEqual(hash(node), hash(Foo()))
+
+    def test_str(self):
+        uri = URI('http://example.com/me/entity#1234')
+        # basic string conversion
+        node = Node(ns.bsfs.Entity, uri)
+        self.assertEqual(str(node), 'Node(http://bsfs.ai/schema/Entity, http://example.com/me/entity#1234)')
+        self.assertEqual(repr(node), 'Node(http://bsfs.ai/schema/Entity, http://example.com/me/entity#1234)')
+        # string conversion respects node_type
+        node = Node(ns.bsfs.Foo, uri)
+        self.assertEqual(str(node), 'Node(http://bsfs.ai/schema/Foo, http://example.com/me/entity#1234)')
+        self.assertEqual(repr(node), 'Node(http://bsfs.ai/schema/Foo, http://example.com/me/entity#1234)')
+        # string conversion respects uri
+        node = Node(ns.bsfs.Entity, URI('http://example.com/me/entity#4321'))
+        self.assertEqual(str(node), 'Node(http://bsfs.ai/schema/Entity, http://example.com/me/entity#4321)')
+        self.assertEqual(repr(node), 'Node(http://bsfs.ai/schema/Entity, http://example.com/me/entity#4321)')
+
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##