Merge branch 'mb/iptc' into develop

5 files changed, 249 insertions, 8 deletions

diff --git a/test/extractor/image/test_iptc.py b/test/extractor/image/test_iptc.py
new file mode 100644
index 0000000..5fa763d
--- /dev/null
+++ b/test/extractor/image/test_iptc.py
@@ -0,0 +1,69 @@
+
+# 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.iptc import Iptc
+
+
+## code ##
+
+class TestIptc(unittest.TestCase):
+
+    def test_eq(self):
+        # identical instances are equal
+        self.assertEqual(Iptc(), Iptc())
+        self.assertEqual(hash(Iptc()), hash(Iptc()))
+        # comparison respects type
+        class Foo(): pass
+        self.assertNotEqual(Iptc(), Foo())
+        self.assertNotEqual(hash(Iptc()), hash(Foo()))
+        self.assertNotEqual(Iptc(), 1234)
+        self.assertNotEqual(hash(Iptc()), hash(1234))
+        self.assertNotEqual(Iptc(), None)
+        self.assertNotEqual(hash(Iptc()), hash(None))
+
+    def test_schema(self):
+        self.assertSetEqual({pred.uri for pred in Iptc().schema.predicates()}, {
+            ns.bsfs.Predicate,
+            ns.bse.tag,
+            ns.bst.label,
+            })
+
+    def test_extract(self):
+        ext = Iptc()
+        node = _node.Node(ns.bsfs.File, '') # Blank node
+        content = {
+            'Iptc.Application2.Keywords': ['hello', 'world'],
+            'Iptc.Application2.RecordVersion': '4',
+            }
+        # target tags
+        t_hello = _node.Node(ns.bsn.Tag, label='hello')
+        t_world = _node.Node(ns.bsn.Tag, label='world')
+
+        # 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.tag)})), {
+            (node, ext.schema.predicate(ns.bse.tag), t_hello),
+            (node, ext.schema.predicate(ns.bse.tag), t_world),
+            (t_hello, ext.schema.predicate(ns.bst.label), 'hello'),
+            (t_world, ext.schema.predicate(ns.bst.label), 'world'),
+            })
+
+        # 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/lib/test_naming_policy.py b/test/lib/test_naming_policy.py
index c9b0cd2..09fd6f6 100644
--- a/test/lib/test_naming_policy.py
+++ b/test/lib/test_naming_policy.py
@@ -23,27 +23,31 @@ class TestDefaultNamingPolicy(unittest.TestCase):
         self.assertEqual(policy.handle_node(
             Node(ns.bsn.Invalid, uri='http://example.com/you/foo#bar')).uri,
             URI('http://example.com/you/foo#bar'))
-        # processes bsfs:File
+        # processes bsn:Entity
         self.assertEqual(policy.handle_node(
             Node(ns.bsn.Entity, ucid='abc123cba')).uri,
             URI('http://example.com/me/file#abc123cba'))
-        # processes bsfs:Preview
+        # processes bsn:Preview
         self.assertEqual(policy.handle_node(
             Node(ns.bsn.Preview, ucid='abc123cba', size=123)).uri,
             URI('http://example.com/me/preview#abc123cba_s123'))
+        # processes bsn:Tag
+        self.assertEqual(policy.handle_node(
+            Node(ns.bsn.Tag, label='hello')).uri,
+            URI('http://example.com/me/tag#hello'))
         # raises an exception on unknown types
         self.assertRaises(errors.ProgrammingError, policy.handle_node,
             Node(ns.bsn.Invalid, ucid='abc123cba', size=123))
 
-    def test_name_file(self):
+    def test_name_entity(self):
         # setup
         policy = DefaultNamingPolicy('http://example.com', 'me')
-        # name_file uses ucid
-        self.assertEqual(policy.name_file(
+        # name_entity uses ucid
+        self.assertEqual(policy.name_entity(
             Node(ns.bsn.Entity, ucid='123abc321')).uri,
             URI('http://example.com/me/file#123abc321'))
-        # name_file falls back to a random guid
-        self.assertTrue(policy.name_file(
+        # name_entity falls back to a random guid
+        self.assertTrue(policy.name_entity(
             Node(ns.bsn.Entity)).uri.startswith('http://example.com/me/file#'))
 
     def test_name_preview(self):
@@ -71,6 +75,29 @@ class TestDefaultNamingPolicy(unittest.TestCase):
         self.assertTrue(policy.name_preview(
             Node(ns.bsn.Preview, size=200)).uri.endswith('_s200'))
 
+    def test_name_tag(self):
+        # setup
+        policy = DefaultNamingPolicy('http://example.com', 'me')
+        # name_tag uses label
+        self.assertEqual(policy.name_tag(
+            Node(ns.bsn.Tag, label='hello')).uri,
+            URI('http://example.com/me/tag#hello'))
+        # name_tag matches the label
+        self.assertEqual(
+            policy.name_tag(Node(ns.bsn.Tag, label='world')),
+            policy.name_tag(Node(ns.bsn.Tag, label='world')),
+            )
+        self.assertNotEqual(
+            policy.name_tag(Node(ns.bsn.Tag, label='hello')),
+            policy.name_tag(Node(ns.bsn.Tag, label='world')),
+            )
+        # label can include characters that are not valid for an uri
+        self.assertEqual(policy.name_tag(
+            Node(ns.bsn.Preview, label='hello world { foo bar ] ')).uri,
+            URI('http://example.com/me/tag#hello%20world%20%7B%20foo%20bar%20%5D%20'))
+        # name_tag falls back to a random guid
+        self.assertTrue(policy.name_tag(
+            Node(ns.bsn.Tag,)).uri.startswith('http://example.com/me/tag#'))
 
 class TestNamingPolicyIterator(unittest.TestCase):
 
diff --git a/test/reader/test_exif.py b/test/reader/test_exif.py
index de6e801..1767f12 100644
--- a/test/reader/test_exif.py
+++ b/test/reader/test_exif.py
@@ -10,7 +10,7 @@ import pyexiv2
 from bsie.utils import errors
 
 # objects to test
-from bsie.reader.exif import Exif
+from bsie.reader.exif import Exif, Iptc
 
 
 ## code ##
@@ -44,6 +44,26 @@ class TestExif(unittest.TestCase):
             })
 
 
+class TestIptc(unittest.TestCase):
+    def test_call(self):
+        rdr = Iptc()
+        # discards non-image files
+        self.assertRaises(errors.UnsupportedFileFormatError, rdr,
+            os.path.join(os.path.dirname(__file__), 'invalid.doc'))
+        # raises on invalid image files
+        self.assertRaises(errors.UnsupportedFileFormatError, rdr,
+            os.path.join(os.path.dirname(__file__), 'invalid.jpg'))
+        # raises on invalid image files
+        pyexiv2.set_log_level(3) # suppress log message
+        self.assertRaises(errors.ReaderError, rdr,
+            os.path.join(os.path.dirname(__file__), 'testimage_exif_corrupted.jpg'))
+        # returns dict with exif info
+        self.assertDictEqual(rdr(os.path.join(os.path.dirname(__file__), 'testimage_exif.jpg')), {
+            'Iptc.Application2.Keywords': ['hello', 'world'],
+            'Iptc.Application2.RecordVersion': '4',
+            })
+
+
 ## main ##
 
 if __name__ == '__main__':
diff --git a/test/reader/testimage_exif.jpg b/test/reader/testimage_exif.jpg
index a774bc2..bc331ac 100644
--- a/test/reader/testimage_exif.jpg
+++ b/test/reader/testimage_exif.jpg
diff --git a/test/utils/test_filewalker.py b/test/utils/test_filewalker.py
new file mode 100644
index 0000000..4aaba65
--- /dev/null
+++ b/test/utils/test_filewalker.py
@@ -0,0 +1,125 @@
+
+# standard imports
+import os
+import shutil
+import tempfile
+import unittest
+
+# objects to test
+from bsie.utils.filewalker import list_files
+
+
+## code ##
+
+def touch(path, text='<test content>'):
+    # create folders
+    os.makedirs(os.path.dirname(path), exist_ok=True)
+    # create file
+    with open(path, 'wt') as ofile:
+        ofile.write(text)
+
+class TestListFiles(unittest.TestCase):
+    def setUp(self):
+        # set up directory structure
+        # <root>
+        # - zero*
+        # - foo
+        #   - hello*
+        #   - remote -> foobar/xyz
+        #   - bar
+        #     - world*
+        #     - xyz*
+        # - bar
+        #   - fst
+        #     - abc*
+        #     - zyx*
+        #   - snd
+        #     - cba*
+        #     - xyz*
+        # - foobar
+        #   - xyz*
+        #   - hello*
+        #   - world -> bar/snd
+        self.testdir = tempfile.mkdtemp(prefix='bsie-test-')
+        touch(os.path.join(self.testdir, 'zero'))
+        touch(os.path.join(self.testdir, 'foo', 'hello'))
+        touch(os.path.join(self.testdir, 'foo', 'bar', 'world'))
+        touch(os.path.join(self.testdir, 'foo', 'bar', 'xyz'))
+        touch(os.path.join(self.testdir, 'bar', 'fst', 'abc'))
+        touch(os.path.join(self.testdir, 'bar', 'fst', 'zyx'))
+        touch(os.path.join(self.testdir, 'bar', 'snd', 'cba'))
+        touch(os.path.join(self.testdir, 'bar', 'snd', 'xyz'))
+        touch(os.path.join(self.testdir, 'foobar', 'xyz'))
+        touch(os.path.join(self.testdir, 'foobar', 'hello'))
+        os.symlink(
+            os.path.join(self.testdir, 'bar', 'snd'),
+            os.path.join(self.testdir, 'foobar', 'world'))
+        os.symlink(
+            os.path.join(self.testdir, 'foobar', 'xyz'),
+            os.path.join(self.testdir, 'foo', 'remote'))
+
+    def tearDown(self):
+        # remove testing dirs
+        shutil.rmtree(self.testdir, ignore_errors=True)
+
+    def test_list_files(self):
+        # list_files lists all files beneath root
+        roots = [
+            os.path.join(self.testdir, 'foo'),
+            os.path.join(self.testdir, 'bar'),
+            os.path.join(self.testdir, 'foobar'),
+            os.path.join(self.testdir, 'zero'),
+            ]
+        self.assertSetEqual(set(list_files(roots, recursive=True, follow_symlinks=True)), {
+            os.path.join(self.testdir, 'bar', 'fst', 'abc'),
+            os.path.join(self.testdir, 'bar', 'fst', 'zyx'),
+            os.path.join(self.testdir, 'bar', 'snd', 'cba'),
+            os.path.join(self.testdir, 'bar', 'snd', 'xyz'),
+            os.path.join(self.testdir, 'foo', 'bar', 'world'),
+            os.path.join(self.testdir, 'foo', 'bar', 'xyz'),
+            os.path.join(self.testdir, 'foo', 'hello'),
+            os.path.join(self.testdir, 'foo', 'remote'),
+            os.path.join(self.testdir, 'foobar', 'hello'),
+            os.path.join(self.testdir, 'foobar', 'world', 'cba'),
+            os.path.join(self.testdir, 'foobar', 'world', 'xyz'),
+            os.path.join(self.testdir, 'foobar', 'xyz'),
+            os.path.join(self.testdir, 'zero'),
+            })
+
+        # list_files lists respects root
+        self.assertSetEqual(set(list_files(
+            roots=[os.path.join(self.testdir, 'foo')], recursive=True, follow_symlinks=True)), {
+            os.path.join(self.testdir, 'foo', 'bar', 'world'),
+            os.path.join(self.testdir, 'foo', 'bar', 'xyz'),
+            os.path.join(self.testdir, 'foo', 'hello'),
+            os.path.join(self.testdir, 'foo', 'remote'),
+            })
+
+        # list_files lists respects recursive flag (lists only files in root!)
+        self.assertSetEqual(set(list_files(roots, recursive=False, follow_symlinks=True)), {
+            os.path.join(self.testdir, 'zero'),
+            })
+
+        # list_files lists respects symlink flag
+        # lists symlinked files but does not dive into symlinked folders
+        self.assertSetEqual(set(list_files(roots, recursive=True, follow_symlinks=False)), {
+            os.path.join(self.testdir, 'bar', 'fst', 'abc'),
+            os.path.join(self.testdir, 'bar', 'fst', 'zyx'),
+            os.path.join(self.testdir, 'bar', 'snd', 'cba'),
+            os.path.join(self.testdir, 'bar', 'snd', 'xyz'),
+            os.path.join(self.testdir, 'foo', 'bar', 'world'),
+            os.path.join(self.testdir, 'foo', 'bar', 'xyz'),
+            os.path.join(self.testdir, 'foo', 'hello'),
+            os.path.join(self.testdir, 'foo', 'remote'),
+            os.path.join(self.testdir, 'foobar', 'hello'),
+            os.path.join(self.testdir, 'foobar', 'xyz'),
+            os.path.join(self.testdir, 'zero'),
+            })
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##