aboutsummaryrefslogtreecommitdiffstats
path: root/test/extractor
diff options
context:
space:
mode:
authorMatthias Baumgartner <dev@igsor.net>2023-03-05 19:22:58 +0100
committerMatthias Baumgartner <dev@igsor.net>2023-03-05 19:22:58 +0100
commita35b33f4f1ddcf6f1bb8ab0f41b87bf2b847f11d (patch)
treefb220da28bb7248ebf37ce09af5de88f2c1aaad4 /test/extractor
parent7582c280ad5324a2f0427999911c7e7abc14a6ab (diff)
parentaf81318ae9311fd0b0e16949cef3cfaf7996970b (diff)
downloadbsie-a35b33f4f1ddcf6f1bb8ab0f41b87bf2b847f11d.tar.gz
bsie-a35b33f4f1ddcf6f1bb8ab0f41b87bf2b847f11d.tar.bz2
bsie-a35b33f4f1ddcf6f1bb8ab0f41b87bf2b847f11d.zip
Merge branch 'develop'HEADv0.23.03releasemain
Diffstat (limited to 'test/extractor')
-rw-r--r--test/extractor/generic/test_constant.py29
-rw-r--r--test/extractor/generic/test_path.py25
-rw-r--r--test/extractor/generic/test_stat.py25
-rw-r--r--test/extractor/image/__init__.py0
-rw-r--r--test/extractor/image/test_colors_spatial.py95
-rw-r--r--test/extractor/image/test_photometrics.py143
-rw-r--r--test/extractor/image/testimage.jpgbin0 -> 349264 bytes
-rw-r--r--test/extractor/test_base.py65
-rw-r--r--test/extractor/test_builder.py98
-rw-r--r--test/extractor/test_preview.py123
-rw-r--r--test/extractor/testimage.jpgbin0 -> 6476 bytes
11 files changed, 556 insertions, 47 deletions
diff --git a/test/extractor/generic/test_constant.py b/test/extractor/generic/test_constant.py
index 9dbaced..77ee02b 100644
--- a/test/extractor/generic/test_constant.py
+++ b/test/extractor/generic/test_constant.py
@@ -1,10 +1,5 @@
-"""
-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
@@ -20,11 +15,11 @@ class TestConstant(unittest.TestCase):
def test_extract(self):
schema = '''
bse:author rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:Entity ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:string ;
bsfs:unique "true"^^xsd:boolean .
bse:comment rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:Entity ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:string ;
bsfs:unique "false"^^xsd:boolean .
'''
@@ -33,33 +28,33 @@ class TestConstant(unittest.TestCase):
(ns.bse.comment, 'the quick brown fox jumps over the lazy dog.'),
]
ext = Constant(schema, tuples)
- node = _node.Node(ns.bsfs.Entity, '') # Blank node
+ node = _node.Node(ns.bsn.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)
+ entity = ext.schema.node(ns.bsfs.Node).child(ns.bsn.Entity)
+ string = ext.schema.literal(ns.bsfs.Literal).child(ns.xsd.string)
# baseline
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
- p_foobar = ext.schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.foobar, domain=entity, range=entity)
+ p_foobar = ext.schema.predicate(ns.bsfs.Predicate).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)
+ p_barfoo = ext.schema.predicate(ns.bse.author).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:domain bsn:Entity ;
rdfs:range xsd:string ;
bsfs:unique "true"^^xsd:boolean .
bse:comment rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:Entity ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:string ;
bsfs:unique "false"^^xsd:boolean .
'''
@@ -80,13 +75,13 @@ class TestConstant(unittest.TestCase):
def test_eq(self):
schema_a = '''
bse:author rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:Entity ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:string ;
bsfs:unique "true"^^xsd:boolean .
'''
schema_b = '''
bse:comment rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:Entity ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:string ;
bsfs:unique "false"^^xsd:boolean .
'''
diff --git a/test/extractor/generic/test_path.py b/test/extractor/generic/test_path.py
index 820f402..0beb37e 100644
--- a/test/extractor/generic/test_path.py
+++ b/test/extractor/generic/test_path.py
@@ -1,14 +1,9 @@
-"""
-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
@@ -29,31 +24,31 @@ class TestPath(unittest.TestCase):
def test_schema(self):
self.assertEqual(Path().schema,
- bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
+ bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
bse:filename rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:File ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:string ;
- bsfs:unique "false"^^xsd:boolean .
+ bsfs:unique "true"^^xsd:boolean .
'''))
def test_extract(self):
ext = Path()
- node = _node.Node(ns.bsfs.File, '') # Blank node
+ node = _node.Node(ns.bsn.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)
+ entity = ext.schema.node(ns.bsfs.Node).child(ns.bsn.Entity)
+ string = ext.schema.literal(ns.bsfs.Literal).child(ns.xsd.string)
# baseline
self.assertSetEqual(set(ext.extract(node, content, (p_filename, ))),
{(node, p_filename, 'bar')})
# predicates parameter is respected
- p_foo = ext.schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.foo, domain=entity, range=string) # unsupported predicate
+ p_foo = ext.schema.predicate(ns.bsfs.Predicate).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
+ p_bar = p_foo.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())
diff --git a/test/extractor/generic/test_stat.py b/test/extractor/generic/test_stat.py
index 3441438..0e83e24 100644
--- a/test/extractor/generic/test_stat.py
+++ b/test/extractor/generic/test_stat.py
@@ -1,15 +1,10 @@
-"""
-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
@@ -30,31 +25,31 @@ class TestStat(unittest.TestCase):
def test_schema(self):
self.assertEqual(Stat().schema,
- bsfs.schema.Schema.from_string(extractor.SCHEMA_PREAMBLE + '''
+ bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
bse:filesize rdfs:subClassOf bsfs:Predicate ;
- rdfs:domain bsfs:File ;
+ rdfs:domain bsn:Entity ;
rdfs:range xsd:integer ;
- bsfs:unique "false"^^xsd:boolean .
+ bsfs:unique "true"^^xsd:boolean .
'''))
def test_extract(self):
ext = Stat()
- node = _node.Node(ns.bsfs.File, '') # Blank node
+ node = _node.Node(ns.bsn.Entity, '') # Blank node
content = os.stat(__file__)
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)
+ entity = ext.schema.node(ns.bsfs.Node).child(ns.bsn.Entity)
+ string = ext.schema.literal(ns.bsfs.Literal).child(ns.xsd.string)
# baseline
self.assertSetEqual(set(ext.extract(node, content, (p_filesize, ))),
{(node, p_filesize, content.st_size)})
# predicates parameter is respected
- p_foo = ext.schema.predicate(ns.bsfs.Predicate).get_child(ns.bse.foo, domain=entity, range=string) # unsupported predicate
+ p_foo = ext.schema.predicate(ns.bsfs.Predicate).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
+ p_bar = p_foo.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())
diff --git a/test/extractor/image/__init__.py b/test/extractor/image/__init__.py
new file mode 100644
index 0000000..e69de29
--- /dev/null
+++ b/test/extractor/image/__init__.py
diff --git a/test/extractor/image/test_colors_spatial.py b/test/extractor/image/test_colors_spatial.py
new file mode 100644
index 0000000..902ab6d
--- /dev/null
+++ b/test/extractor/image/test_colors_spatial.py
@@ -0,0 +1,95 @@
+
+# standard imports
+import os
+import unittest
+
+# external imports
+import PIL.Image
+
+# bsie imports
+from bsie.extractor import base
+from bsie.utils import bsfs, ns, node as _node
+
+# objects to test
+from bsie.extractor.image.colors_spatial import ColorsSpatial
+
+
+## code ##
+
+class TestColorsSpatial(unittest.TestCase):
+ def setUp(self):
+ # content id with default constructors (width=32, height=32, exp=4)
+ self.instance_prefix = 'https://schema.bsfs.io/ie/Literal/Array/Feature/ColorsSpatial'
+ self.predicate_prefix = 'https://schema.bsfs.io/ie/Node/Entity#colors_spatial_'
+ self.uuid = 'adee8d6c43687021e1c5bffe56bcfe727f1638d792744137181304ef889dac2a'
+
+ def test_essentials(self):
+ # clones are equal
+ self.assertEqual(ColorsSpatial(32, 32, 4), ColorsSpatial(32, 32, 4))
+ self.assertEqual(hash(ColorsSpatial(32, 32, 4)), hash(ColorsSpatial(32, 32, 4)))
+ # equal respects type
+ self.assertNotEqual(ColorsSpatial(32, 32, 4), 'hello world')
+ self.assertNotEqual(hash(ColorsSpatial(32, 32, 4)), hash('hello world'))
+ # equals respects width
+ self.assertNotEqual(ColorsSpatial(32, 32, 4), ColorsSpatial(16, 32, 4))
+ self.assertNotEqual(hash(ColorsSpatial(32, 32, 4)), hash(ColorsSpatial(16, 32, 4)))
+ # equals respects height
+ self.assertNotEqual(ColorsSpatial(32, 32, 4), ColorsSpatial(32, 16, 4))
+ self.assertNotEqual(hash(ColorsSpatial(32, 32, 4)), hash(ColorsSpatial(32, 16, 4)))
+ # equals respects exp
+ self.assertNotEqual(ColorsSpatial(32, 32, 4), ColorsSpatial(32, 32, 8))
+ self.assertNotEqual(hash(ColorsSpatial(32, 32, 4)), hash(ColorsSpatial(32, 32, 8)))
+ # string representation
+ self.assertEqual(str(ColorsSpatial()), 'ColorsSpatial')
+ self.assertEqual(repr(ColorsSpatial(64, 16, 2)), 'ColorsSpatial(64, 16, 2)')
+
+ def test_dimension(self):
+ self.assertEqual(ColorsSpatial.dimension(32, 32, 4), 3 * (32*32 + 8*8 + 2*2))
+ self.assertEqual(ColorsSpatial.dimension(16, 16, 8), 3 * (16*16 + 2*2))
+ self.assertEqual(ColorsSpatial.dimension(64, 64, 16), 3 * (64*64 + 4*4))
+
+ def test_schema(self):
+ schema = bsfs.schema.from_string(base.SCHEMA_PREAMBLE + f'''
+ <{self.instance_prefix}> rdfs:subClassOf bsa:Feature ;
+ # annotations
+ rdfs:label "Spatially dominant colors"^^xsd:string ;
+ schema:description "Domiant colors of subregions in an image."^^xsd:string ;
+ bsfs:dtype xsd:integer .
+
+ <{self.instance_prefix}#{self.uuid}> rdfs:subClassOf <{self.instance_prefix}> ;
+ bsfs:dimension "3276"^^xsd:integer ;
+ # annotations
+ <{self.instance_prefix}/args#width> "32"^^xsd:integer ;
+ <{self.instance_prefix}/args#height> "32"^^xsd:integer ;
+ <{self.instance_prefix}/args#exp> "4"^^xsd:float .
+
+ <{self.predicate_prefix}{self.uuid}> rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Entity ;
+ rdfs:range <{self.instance_prefix}#{self.uuid}> ;
+ bsfs:unique "true"^^xsd:boolean .
+ ''')
+ self.assertEqual(schema, ColorsSpatial().schema)
+
+ def test_extract(self):
+ ext = ColorsSpatial(2,2,2)
+ img = PIL.Image.open(os.path.join(os.path.dirname(__file__), 'testimage.jpg'))
+ node = _node.Node(ns.bsn.Entity, bsfs.URI('http://example.com/entity#1234'))
+ principals = set(ext.principals)
+ self.assertEqual(len(principals), 1)
+ # valid invocation yields feature
+ ret = list(ext.extract(node, img, principals))
+ self.assertEqual(ret[0], (
+ node,
+ list(principals)[0],
+ (91, 127, 121, 94, 138, 167, 163, 134, 190, 138, 170, 156, 121, 142, 159)))
+ # principals is respected
+ self.assertListEqual(list(ext.extract(node, img, {})), [])
+
+
+
+## main ##
+
+if __name__ == '__main__':
+ unittest.main()
+
+## EOF ##
diff --git a/test/extractor/image/test_photometrics.py b/test/extractor/image/test_photometrics.py
new file mode 100644
index 0000000..fb219e2
--- /dev/null
+++ b/test/extractor/image/test_photometrics.py
@@ -0,0 +1,143 @@
+
+# standard imports
+import unittest
+
+# bsie imports
+from bsie.extractor import base
+from bsie.utils import bsfs, node as _node, ns
+
+# objects to test
+from bsie.extractor.image.photometrics import Exif, _gps_to_dec
+
+
+## code ##
+
+class TestExif(unittest.TestCase):
+
+ def test_gps_to_dec(self):
+ # deg+min+sec format
+ self.assertAlmostEqual(_gps_to_dec('29/1 58/1 45/1'.split()), 29.979167, 6)
+ self.assertAlmostEqual(_gps_to_dec('31 08 03'.split()), 31.134167, 6)
+ self.assertAlmostEqual(_gps_to_dec('20 40 586/10'.split()), 20.682944, 6)
+ self.assertAlmostEqual(_gps_to_dec('88/1 34 68/10'.split()), 88.568556, 6)
+ # deg+min format
+ self.assertAlmostEqual(_gps_to_dec('13 472167/10000 0/1 '.split()), 13.786945, 6)
+ self.assertAlmostEqual(_gps_to_dec('104/1 3215/100 0/1'.split()), 104.535833, 6)
+
+ def test_eq(self):
+ # identical instances are equal
+ self.assertEqual(Exif(), Exif())
+ self.assertEqual(hash(Exif()), hash(Exif()))
+ # comparison respects type
+ class Foo(): pass
+ self.assertNotEqual(Exif(), Foo())
+ self.assertNotEqual(hash(Exif()), hash(Foo()))
+ self.assertNotEqual(Exif(), 1234)
+ self.assertNotEqual(hash(Exif()), hash(1234))
+ self.assertNotEqual(Exif(), None)
+ self.assertNotEqual(hash(Exif()), hash(None))
+
+ def test_schema(self):
+ self.assertSetEqual({pred.uri for pred in Exif().schema.predicates()}, {
+ ns.bsfs.Predicate,
+ ns.bse.exposure,
+ ns.bse.aperture,
+ ns.bse.iso,
+ ns.bse.focal_length,
+ ns.bse.width,
+ ns.bse.height,
+ ns.bse.orientation,
+ ns.bse.orientation_label,
+ ns.bse.altitude,
+ ns.bse.latitude,
+ ns.bse.longitude,
+ })
+
+ def test_extract(self):
+ ext = Exif()
+ node = _node.Node(ns.bsfs.File, '') # Blank node
+ content = {
+ 'Exif.Photo.ExposureTime': '10/600',
+ 'Exif.Photo.FNumber': '48/10',
+ 'Exif.Photo.ISOSpeedRatings': '400',
+ 'Exif.Photo.FocalLength': '460/10',
+ 'Exif.Photo.PixelXDimension': '4288',
+ 'Exif.Photo.PixelYDimension': '2848',
+ 'Exif.Image.Orientation': '1',
+ 'Exif.GPSInfo.GPSAltitude': '431/1',
+ 'Exif.GPSInfo.GPSLatitude': '46/1 11397/625 0/1',
+ 'Exif.GPSInfo.GPSLongitude': '7/1 131250/2500 0/1',
+ }
+
+ # invalid principals are ignored
+ self.assertSetEqual(set(ext.extract(node, content, {ns.bse.filename})), set())
+ # extract finds all relevant information
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.exposure)})),
+ {(node, ext.schema.predicate(ns.bse.exposure), 60.0)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.aperture)})),
+ {(node, ext.schema.predicate(ns.bse.aperture), 4.8)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.iso)})),
+ {(node, ext.schema.predicate(ns.bse.iso), 400)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.focal_length)})),
+ {(node, ext.schema.predicate(ns.bse.focal_length), 46.0)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.width)})),
+ {(node, ext.schema.predicate(ns.bse.width), 4288)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.height)})),
+ {(node, ext.schema.predicate(ns.bse.height), 2848)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.orientation)})),
+ {(node, ext.schema.predicate(ns.bse.orientation), 1)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.orientation_label)})),
+ {(node, ext.schema.predicate(ns.bse.orientation_label), 'landscape')})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.altitude)})),
+ {(node, ext.schema.predicate(ns.bse.altitude), 431.0)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.latitude)})),
+ {(node, ext.schema.predicate(ns.bse.latitude), 46.30392)})
+ self.assertSetEqual(set(ext.extract(node, content, {ext.schema.predicate(ns.bse.longitude)})),
+ {(node, ext.schema.predicate(ns.bse.longitude), 7.875)})
+
+ # check orientation label
+ self.assertSetEqual(set(ext.extract(
+ node, {
+ 'Exif.Photo.PixelXDimension': '4288',
+ 'Exif.Photo.PixelYDimension': '2848',
+ 'Exif.Image.Orientation': '5',
+ },
+ {ext.schema.predicate(ns.bse.orientation_label)})),
+ {(node, ext.schema.predicate(ns.bse.orientation_label), 'portrait')})
+
+ # can pass multiple principals
+ self.assertSetEqual(set(ext.extract(node, content, {
+ ext.schema.predicate(ns.bse.exposure),
+ ext.schema.predicate(ns.bse.iso),
+ ext.schema.predicate(ns.bse.focal_length),
+ })), {
+ (node, ext.schema.predicate(ns.bse.exposure), 60.0),
+ (node, ext.schema.predicate(ns.bse.iso), 400),
+ (node, ext.schema.predicate(ns.bse.focal_length), 46.0),
+ })
+
+ # principals w/o content are ignored
+ self.assertSetEqual(set(ext.extract(
+ node,
+ content={'Exif.Photo.ExposureTime': '10/600'},
+ principals={
+ ext.schema.predicate(ns.bse.exposure),
+ ext.schema.predicate(ns.bse.iso),
+ ext.schema.predicate(ns.bse.focal_length),
+ })
+ ), {
+ (node, ext.schema.predicate(ns.bse.exposure), 60.0),
+ })
+
+ # empty content is acceptable
+ self.assertSetEqual(set(ext.extract(node, {}, set(ext.principals))), set())
+ # no principals is acceptable
+ self.assertSetEqual(set(ext.extract(node, content, set())), set())
+
+
+## main ##
+
+if __name__ == '__main__':
+ unittest.main()
+
+## EOF ##
diff --git a/test/extractor/image/testimage.jpg b/test/extractor/image/testimage.jpg
new file mode 100644
index 0000000..c80bb48
--- /dev/null
+++ b/test/extractor/image/testimage.jpg
Binary files differ
diff --git a/test/extractor/test_base.py b/test/extractor/test_base.py
new file mode 100644
index 0000000..81865e1
--- /dev/null
+++ b/test/extractor/test_base.py
@@ -0,0 +1,65 @@
+
+# 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.from_string(base.SCHEMA_PREAMBLE + '''
+ bse:author rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Entity ;
+ rdfs:range xsd:string ;
+ bsfs:unique "false"^^xsd:boolean .
+ bse:comment rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn: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()
+ entity = schema.node(ns.bsfs.Node).child(ns.bsn.Entity)
+ string = schema.literal(ns.bsfs.Literal).child(ns.xsd.string)
+ p_author = schema.predicate(ns.bsfs.Predicate).child(ns.bse.author, domain=entity, range=string)
+ p_comment = schema.predicate(ns.bsfs.Predicate).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..fbb0895
--- /dev/null
+++ b/test/extractor/test_builder.py
@@ -0,0 +1,98 @@
+
+# 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 bsn:Entity ;
+ rdfs:range xsd:string ;
+ bsfs:unique "true"^^xsd:boolean .
+ bse:rating rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Entity ;
+ rdfs:range xsd:integer ;
+ bsfs:unique "true"^^xsd:boolean .
+ ''',
+ 'tuples': [
+ ('https://schema.bsfs.io/ie/Node/Entity#author', 'Me, myself, and I'),
+ ('https://schema.bsfs.io/ie/Node/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 bsn:Entity ;
+ rdfs:range xsd:string ;
+ bsfs:unique "true"^^xsd:boolean .
+ bse:rating rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Entity ;
+ rdfs:range xsd:integer ;
+ bsfs:unique "true"^^xsd:boolean .
+ ''', [
+ ('https://schema.bsfs.io/ie/Node/Entity#author', 'Me, myself, and I'),
+ ('https://schema.bsfs.io/ie/Node/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 ##
diff --git a/test/extractor/test_preview.py b/test/extractor/test_preview.py
new file mode 100644
index 0000000..6526783
--- /dev/null
+++ b/test/extractor/test_preview.py
@@ -0,0 +1,123 @@
+
+# standard imports
+import io
+import os
+import unittest
+
+# external imports
+import PIL.Image
+
+# bsie imports
+from bsie.extractor import base
+from bsie.utils import bsfs, node as _node, ns
+from bsie.reader.preview import Preview as Reader
+
+# objects to test
+from bsie.extractor.preview import Preview
+
+
+## code ##
+
+class TestPreview(unittest.TestCase):
+ def test_eq(self):
+ # identical instances are equal
+ self.assertEqual(Preview([1,2,3]), Preview([1,2,3]))
+ self.assertEqual(hash(Preview([1,2,3])), hash(Preview([1,2,3])))
+ # comparison respects max_sides
+ self.assertNotEqual(Preview([1,2,3]), Preview([1,2]))
+ self.assertNotEqual(hash(Preview([1,2,3])), hash(Preview([1,2])))
+ self.assertNotEqual(Preview([1,2]), Preview([1,2,3]))
+ self.assertNotEqual(hash(Preview([1,2])), hash(Preview([1,2,3])))
+ # comparison respects type
+ class Foo(): pass
+ self.assertNotEqual(Preview([1,2,3]), Foo())
+ self.assertNotEqual(hash(Preview([1,2,3])), hash(Foo()))
+ self.assertNotEqual(Preview([1,2,3]), 123)
+ self.assertNotEqual(hash(Preview([1,2,3])), hash(123))
+ self.assertNotEqual(Preview([1,2,3]), None)
+ self.assertNotEqual(hash(Preview([1,2,3])), hash(None))
+
+ def test_schema(self):
+ self.assertEqual(Preview([1,2,3]).schema,
+ bsfs.schema.from_string(base.SCHEMA_PREAMBLE + '''
+ bsn:Preview rdfs:subClassOf bsfs:Node .
+ bsl:BinaryBlob rdfs:subClassOf bsfs:Literal .
+ <https://schema.bsfs.io/ie/Literal/BinaryBlob/JPEG> rdfs:subClassOf bsl:BinaryBlob .
+
+ bse:preview rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Entity ;
+ rdfs:range bsn:Preview ;
+ bsfs:unique "false"^^xsd:boolean .
+
+ bsp:width rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Preview ;
+ rdfs:range xsd:integer ;
+ bsfs:unique "true"^^xsd:boolean .
+
+ bsp:height rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Preview ;
+ rdfs:range xsd:integer ;
+ bsfs:unique "true"^^xsd:boolean .
+
+ bsp:asset rdfs:subClassOf bsfs:Predicate ;
+ rdfs:domain bsn:Preview ;
+ rdfs:range <https://schema.bsfs.io/ie/Literal/BinaryBlob/JPEG> ;
+ bsfs:unique "true"^^xsd:boolean .
+
+ '''))
+
+ def test_extract(self):
+ # setup dependents
+ rdr = Reader()
+ subject = _node.Node(ns.bsn.Entity)
+ path = os.path.join(os.path.dirname(__file__), 'testimage.jpg')
+
+ # setup extractor
+ ext = Preview(max_sides=[10])
+ principals = set(ext.principals)
+ self.assertEqual(principals, {ext.schema.predicate(ns.bse.preview)})
+ # skip unknown predicates
+ gen = rdr(path)
+ self.assertSetEqual(set(), set(ext.extract(subject, gen,
+ {ext.schema.predicate(ns.bsfs.Predicate).child(ns.bse.unknown)})))
+ gen(10) # NOTE: consume some image to avoid resource error warning
+ # extract a preview
+ triples = set(ext.extract(subject, rdr(path), principals))
+ thumbs = {node for node, _, _ in triples if node.node_type == ns.bsn.Preview}
+ self.assertEqual(len(thumbs), 1)
+ thumb = list(thumbs)[0]
+ # test properties
+ self.assertTrue(triples.issuperset({
+ (subject, ext.schema.predicate(ns.bse.preview), thumb),
+ (thumb, ext.schema.predicate(ns.bsp.width), 10),
+ (thumb, ext.schema.predicate(ns.bsp.height), 10),
+ }))
+ # test image data
+ rawdata = {val for _, pred, val in triples if pred == ext.schema.predicate(ns.bsp.asset)}
+ self.assertEqual(len(rawdata), 1)
+ data = io.BytesIO(list(rawdata)[0])
+ data.seek(0)
+ img = PIL.Image.open(data)
+ self.assertEqual(img.size, (10, 10))
+ self.assertEqual(sum(band for pix in img.getdata() for band in pix), 0)
+
+ # setup extractor
+ ext = Preview(max_sides=[10, 20])
+ principals = set(ext.principals)
+ self.assertEqual(principals, {ext.schema.predicate(ns.bse.preview)})
+ # extract a preview
+ triples = set(ext.extract(subject, rdr(path), principals))
+ thumbs = {node for node, _, _ in triples if node.node_type == ns.bsn.Preview}
+ self.assertEqual(len(thumbs), 2)
+ self.assertSetEqual({10, 20}, {
+ value for _, pred, value in triples if pred == ext.schema.predicate(ns.bsp.width)})
+ self.assertSetEqual({10, 20}, {
+ value for _, pred, value in triples if pred == ext.schema.predicate(ns.bsp.height)})
+
+
+## main ##
+
+if __name__ == '__main__':
+ unittest.main()
+
+## EOF ##
diff --git a/test/extractor/testimage.jpg b/test/extractor/testimage.jpg
new file mode 100644
index 0000000..4c2aca5
--- /dev/null
+++ b/test/extractor/testimage.jpg
Binary files differ