bsie/pipeline interface revision:

* predicates -> principals
* schema as property
* principals as property
* information hiding
* full subschema instead of only predicates

2 files changed, 13 insertions, 66 deletions

diff --git a/test/lib/test_bsie.py b/test/lib/test_bsie.py
index 43e7b1d..f3f476e 100644
--- a/test/lib/test_bsie.py
+++ b/test/lib/test_bsie.py
@@ -11,8 +11,6 @@ import unittest
 # bsie imports
 from bsie.base import extractor
 from bsie.tools import builder
-from bsie.utils.bsfs import URI, schema
-from bsie.utils.node import Node
 from bsie.utils import bsfs, node, ns
 
 # objects to test
@@ -47,22 +45,12 @@ class TestBSIE(unittest.TestCase):
     def test_construction(self):
         # pipeline only
         lib = BSIE(self.pipeline)
-        self.assertSetEqual(lib.predicates, {
+        self.assertSetEqual(set(lib.principals), {
             ns.bse.filename,
             ns.bse.filesize,
             ns.bse.author,
             })
-        self.assertEqual(lib.schema, schema.Schema.from_string('''
-            prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
-            prefix xsd: <http://www.w3.org/2001/XMLSchema#>
-            prefix bsfs: <http://bsfs.ai/schema/>
-            prefix bse: <http://bsfs.ai/schema/Entity#>
-            # essential nodes
-            bsfs:Entity rdfs:subClassOf bsfs:Node .
-            # common definitions
-            xsd:string rdfs:subClassOf bsfs:Literal .
-            xsd:integer rdfs:subClassOf bsfs:Literal .
-
+        self.assertEqual(lib.schema, bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
             bse:filename rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:File ;
                 rdfs:range xsd:string ;
@@ -85,21 +73,11 @@ class TestBSIE(unittest.TestCase):
             ns.bse.author,
             ns.bse.inexistent,
             })
-        self.assertSetEqual(lib.predicates, {
+        self.assertSetEqual(set(lib.principals), {
             ns.bse.filesize,
             ns.bse.author,
             })
-        self.assertEqual(lib.schema, schema.Schema.from_string('''
-            prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
-            prefix xsd: <http://www.w3.org/2001/XMLSchema#>
-            prefix bsfs: <http://bsfs.ai/schema/>
-            prefix bse: <http://bsfs.ai/schema/Entity#>
-            # essential nodes
-            bsfs:Entity rdfs:subClassOf bsfs:Node .
-            # common definitions
-            xsd:string rdfs:subClassOf bsfs:Literal .
-            xsd:integer rdfs:subClassOf bsfs:Literal .
-
+        self.assertEqual(lib.schema, bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
             bse:filesize rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:File ;
                 rdfs:range xsd:integer;
@@ -109,26 +87,15 @@ class TestBSIE(unittest.TestCase):
                 rdfs:domain bsfs:Entity ;
                 rdfs:range xsd:string ;
                 bsfs:unique "true"^^xsd:boolean .
-
             '''))
         # empty collect is disregarded
         lib = BSIE(self.pipeline, collect={})
-        self.assertSetEqual(lib.predicates, {
+        self.assertSetEqual(set(lib.principals), {
             ns.bse.filename,
             ns.bse.filesize,
             ns.bse.author,
             })
-        self.assertEqual(lib.schema, schema.Schema.from_string('''
-            prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
-            prefix xsd: <http://www.w3.org/2001/XMLSchema#>
-            prefix bsfs: <http://bsfs.ai/schema/>
-            prefix bse: <http://bsfs.ai/schema/Entity#>
-            # essential nodes
-            bsfs:Entity rdfs:subClassOf bsfs:Node .
-            # common definitions
-            xsd:string rdfs:subClassOf bsfs:Literal .
-            xsd:integer rdfs:subClassOf bsfs:Literal .
-
+        self.assertEqual(lib.schema, bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
             bse:filename rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:File ;
                 rdfs:range xsd:string ;
@@ -152,24 +119,14 @@ class TestBSIE(unittest.TestCase):
             ns.bse.filename,
             ns.bse.inexistent,
             })
-        self.assertSetEqual(lib.predicates, {
+        self.assertSetEqual(set(lib.principals), {
             ns.bse.author,
             })
-        self.assertEqual(lib.schema, schema.Schema.from_string('''
-            prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
-            prefix xsd: <http://www.w3.org/2001/XMLSchema#>
-            prefix bsfs: <http://bsfs.ai/schema/>
-            prefix bse: <http://bsfs.ai/schema/Entity#>
-            # essential nodes
-            bsfs:Entity rdfs:subClassOf bsfs:Node .
-            # common definitions
-            xsd:string rdfs:subClassOf bsfs:Literal .
-
+        self.assertEqual(lib.schema, bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
             bse:author rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:Entity ;
                 rdfs:range xsd:string ;
                 bsfs:unique "true"^^xsd:boolean .
-
             '''))
 
         # specify collect and discard
@@ -177,19 +134,10 @@ class TestBSIE(unittest.TestCase):
             collect={ns.bse.filesize, ns.bse.author, ns.bse.foo, ns.bse.bar},
             discard={ns.bse.author, ns.bse.foo, ns.bse.foobar},
             )
-        self.assertSetEqual(lib.predicates, {
+        self.assertSetEqual(set(lib.principals), {
             ns.bse.filesize,
             })
-        self.assertEqual(lib.schema, schema.Schema.from_string('''
-            prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>
-            prefix xsd: <http://www.w3.org/2001/XMLSchema#>
-            prefix bsfs: <http://bsfs.ai/schema/>
-            prefix bse: <http://bsfs.ai/schema/Entity#>
-            # essential nodes
-            bsfs:Entity rdfs:subClassOf bsfs:Node .
-            # common definitions
-            xsd:integer rdfs:subClassOf bsfs:Literal .
-
+        self.assertEqual(lib.schema, bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
             bse:filesize rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:File ;
                 rdfs:range xsd:integer;
@@ -201,7 +149,7 @@ class TestBSIE(unittest.TestCase):
     def test_from_file(self):
         # setup
         lib = BSIE(self.pipeline)
-        self.assertSetEqual(set(lib.predicates), {
+        self.assertSetEqual(set(lib.principals), {
             ns.bse.filesize,
             ns.bse.filename,
             ns.bse.author,
diff --git a/test/tools/test_pipeline.py b/test/tools/test_pipeline.py
index e440ab5..91bf736 100644
--- a/test/tools/test_pipeline.py
+++ b/test/tools/test_pipeline.py
@@ -75,7 +75,7 @@ class TestPipeline(unittest.TestCase):
 
         # equivalence respects schema
         p2 = Pipeline(self.prefix, self.ext2rdr)
-        p2.schema = pipeline.schema.Empty()
+        p2._schema = pipeline.schema.Empty()
         self.assertNotEqual(pipeline, p2)
         self.assertNotEqual(hash(pipeline), hash(p2))
 
@@ -160,8 +160,7 @@ class TestPipeline(unittest.TestCase):
         # build pipeline
         pipeline = Pipeline(self.prefix, self.ext2rdr)
         #
-        self.assertSetEqual(set(pipeline.predicates()), {
-            pipeline.schema.predicate(ns.bsfs.Predicate),
+        self.assertSetEqual(set(pipeline.principals), {
             pipeline.schema.predicate(ns.bse.filename),
             pipeline.schema.predicate(ns.bse.filesize),
             pipeline.schema.predicate(ns.bse.author),