Merge branch 'mb/feature' into develop

4 files changed, 0 insertions, 423 deletions

diff --git a/test/tools/__init__.py b/test/tools/__init__.py
deleted file mode 100644
index e69de29..0000000
--- a/test/tools/__init__.py
+++ /dev/null
diff --git a/test/tools/test_builder.py b/test/tools/test_builder.py
deleted file mode 100644
index 62c637c..0000000
--- a/test/tools/test_builder.py
+++ /dev/null
@@ -1,246 +0,0 @@
-"""
-
-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.utils import bsfs
-
-# 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(base.errors.LoaderError, _safe_load, 'dBGHMSAYOoKeKMpywDoKZQycENFPvN', 'foobar')
-        self.assertRaises(base.errors.LoaderError, _safe_load, 'dBGHMSAYOoKeKMpywDoKZQycENFPvN.bar', 'foobar')
-        # partially valid module
-        self.assertRaises(base.errors.LoaderError, _safe_load, 'os.foo', 'foobar')
-        # invalid class
-        self.assertRaises(base.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(base.errors.BuilderError, builder.build, 'path.Path')
-        # invalid config
-        builder = ReaderBuilder({'bsie.reader.stat.Stat': dict(foo=123)})
-        self.assertRaises(base.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 ;
-                        bsfs:unique "true"^^xsd:boolean .
-                    bse:rating rdfs:subClassOf bsfs:Predicate ;
-                        rdfs:domain bsfs:Entity ;
-                        rdfs:range xsd:integer ;
-                        bsfs:unique "true"^^xsd:boolean .
-                    ''',
-                '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 ;
-                        bsfs:unique "true"^^xsd:boolean .
-                    bse:rating rdfs:subClassOf bsfs:Predicate ;
-                        rdfs:domain bsfs:Entity ;
-                        rdfs:range xsd:integer ;
-                        bsfs:unique "true"^^xsd:boolean .
-                    ''', [
-            ('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I'),
-            ('http://bsfs.ai/schema/Entity#rating', 123),
-            ]))
-
-        # building with invalid args
-        self.assertRaises(base.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 = bsfs.URI('http://example.com/local/file#')
-        c_schema = '''
-            bse:author rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:Entity ;
-                rdfs:range xsd:string ;
-                bsfs:unique "true"^^xsd:boolean .
-            '''
-        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
deleted file mode 100644
index a116a30..0000000
--- a/test/tools/test_pipeline.py
+++ /dev/null
@@ -1,176 +0,0 @@
-"""
-
-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 bsfs, node, ns
-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:File ;
-                rdfs:range xsd:string ;
-                bsfs:unique "true"^^xsd:boolean .
-            '''
-        tupA = [('http://bsfs.ai/schema/Entity#author', 'Me, myself, and I')]
-        # constant B
-        csB = '''
-            bse:rating rdfs:subClassOf bsfs:Predicate ;
-                rdfs:domain bsfs:File ;
-                rdfs:range xsd:integer ;
-                bsfs:unique "true"^^xsd:boolean .
-            '''
-        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 = bsfs.Namespace('http://example.com/local/')
-
-    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(bsfs.URI('http://example.com/global/ent#'), self.ext2rdr))
-        self.assertNotEqual(hash(pipeline), hash(Pipeline(bsfs.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 = 'a948904f2f0f479b8f8197694b30184b0d2ed1c1cd2a1ec0fb85d299a192a447'
-        subject = node.Node(ns.bsfs.File, (self.prefix + 'file#')[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.File)
-        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, 12),
-            })
-        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, 12),
-            })
-        # extract all predicates
-        self.assertSetEqual(set(pipeline(testfile)), {
-            (subject, p_filename, 'testfile.t'),
-            (subject, p_filesize, 12),
-            (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())
-
-    def test_predicates(self):
-        # build pipeline
-        pipeline = Pipeline(self.prefix, self.ext2rdr)
-        #
-        self.assertSetEqual(set(pipeline.principals), {
-            pipeline.schema.predicate(ns.bse.filename),
-            pipeline.schema.predicate(ns.bse.filesize),
-            pipeline.schema.predicate(ns.bse.author),
-            pipeline.schema.predicate(ns.bse.rating),
-            })
-
-
-## main ##
-
-if __name__ == '__main__':
-    unittest.main()
-
-## EOF ##
diff --git a/test/tools/testfile.t b/test/tools/testfile.t
deleted file mode 100644
index 3b18e51..0000000
--- a/test/tools/testfile.t
+++ /dev/null
@@ -1 +0,0 @@
-hello world