package restructuring: base

* Reader and Extractor to respective reader/extractor modules
* ReaderBuilder to reader module
* ExtractorBuilder to extractor module
* Loading module in utils (safe_load, unpack_name)
* Pipeline and PipelineBuilder to lib module
* errors to utils
* documentation: "standard import" and "external import"

4 files changed, 179 insertions, 6 deletions

diff --git a/test/extractor/generic/test_path.py b/test/extractor/generic/test_path.py
index 820f402..778ac5a 100644
--- a/test/extractor/generic/test_path.py
+++ b/test/extractor/generic/test_path.py
@@ -4,11 +4,11 @@ Part of the bsie test suite.
 A copy of the license is provided with the project.
 Author: Matthias Baumgartner, 2022
 """
-# imports
+# standard imports
 import unittest
 
 # bsie imports
-from bsie.base import extractor
+from bsie.extractor import base
 from bsie.utils import bsfs, node as _node, ns
 
 # objects to test
@@ -29,7 +29,7 @@ class TestPath(unittest.TestCase):
 
     def test_schema(self):
         self.assertEqual(Path().schema,
-            bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
+            bsfs.schema.Schema.from_string(base.SCHEMA_PREAMBLE + '''
                 bse:filename rdfs:subClassOf bsfs:Predicate ;
                     rdfs:domain bsfs:File ;
                     rdfs:range xsd:string ;
diff --git a/test/extractor/generic/test_stat.py b/test/extractor/generic/test_stat.py
index 3441438..ff74085 100644
--- a/test/extractor/generic/test_stat.py
+++ b/test/extractor/generic/test_stat.py
@@ -4,12 +4,12 @@ Part of the bsie test suite.
 A copy of the license is provided with the project.
 Author: Matthias Baumgartner, 2022
 """
-# imports
+# standard imports
 import os
 import unittest
 
 # bsie imports
-from bsie.base import extractor
+from bsie.extractor import base
 from bsie.utils import bsfs, node as _node, ns
 
 # objects to test
@@ -30,7 +30,7 @@ class TestStat(unittest.TestCase):
 
     def test_schema(self):
         self.assertEqual(Stat().schema,
-            bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
+            bsfs.schema.Schema.from_string(base.SCHEMA_PREAMBLE + '''
                 bse:filesize rdfs:subClassOf bsfs:Predicate ;
                     rdfs:domain bsfs:File ;
                     rdfs:range xsd:integer ;
diff --git a/test/extractor/test_base.py b/test/extractor/test_base.py
new file mode 100644
index 0000000..6a63c59
--- /dev/null
+++ b/test/extractor/test_base.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
+"""
+# standard imports
+import unittest
+
+# bsie imports
+from bsie.utils import bsfs, ns
+
+# objects to test
+from bsie.extractor import base
+
+
+## code ##
+
+class StubExtractor(base.Extractor):
+    def __init__(self):
+        super().__init__(bsfs.schema.Schema.from_string(base.SCHEMA_PREAMBLE + '''
+            bse:author rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                bsfs:unique "false"^^xsd:boolean .
+            bse:comment rdfs:subClassOf bsfs:Predicate ;
+                rdfs:domain bsfs:Entity ;
+                rdfs:range xsd:string ;
+                bsfs:unique "false"^^xsd:boolean .
+        '''))
+
+    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_principals(self):
+        schema = bsfs.schema.Schema.Empty()
+        entity = schema.node(ns.bsfs.Node).get_child(ns.bsfs.Entity)
+        string = schema.literal(ns.bsfs.Literal).get_child(bsfs.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.principals),
+            {p_author, p_comment} | set(schema.predicates()) - {schema.predicate(ns.bsfs.Predicate)})
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/extractor/test_builder.py b/test/extractor/test_builder.py
new file mode 100644
index 0000000..039ea53
--- /dev/null
+++ b/test/extractor/test_builder.py
@@ -0,0 +1,103 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# standard imports
+import unittest
+
+# bsie imports
+from bsie.utils import errors
+
+# objects to test
+from bsie.extractor import ExtractorBuilder
+
+
+## code ##
+
+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(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)
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##