def load_meta(img_file: Path) -> ImageMetadata:
"""
Load the metadata tuple for a given image file.
If no metadata file is present, or it is currently inaccessible, return a blank metadata tuple.
:arg img_file: a path pointing to a managed image for which we want to load metadata
:return: the associated metadata as a tuple, or a blank metadata tuple
"""
meta_file = _construct_metadata_path(img_file)
if meta_file.exists():
try:
with meta_file.open() as mf:
metadata = parse_xml(mf.read())
# Check if 'file' is a valid URI, otherwise make it so (for retro-compatibility with older schema)
if metadata.file.scheme is None:
metadata = _old_to_new_schema(img_file, metadata)
except (OSError, ParseError):
metadata = ImageMetadata(uuid3(NAMESPACE_URL, str(URI(img_file))),
URI(img_file), None, None, None, None)
else:
metadata = ImageMetadata(uuid3(NAMESPACE_URL, str(URI(img_file))),
URI(img_file), None, None, None, None)
return metadata
def test_collective_disjunctive_filter(self):
# Verify the effectiveness of multi-valued matchers, when evaluated in disjunction
chars1 = ["al", "john", "jack"]
chars2 = ["jm", "jr"]
chars3 = ["jr"]
el1 = ImageMetadata(uuid4(), URI("a.png"), "ghi", None, chars1, None)
el2 = ImageMetadata(uuid4(), URI("b.png"), "nsh", None, chars2, None)
el3 = ImageMetadata(uuid4(), URI("c.png"), "ShT", None, chars3, None)
filter_builder = FilterBuilder()
# Test disjunctive filtering with inclusion
f = filter_builder.character_constraint("jm").character_constraint("al").characters_as_disjunctive(True) \
.get_character_filter()
self.assertTrue(f(el1))
self.assertTrue(f(el2))
self.assertFalse(f(el3))
# Test disjunctive filtering with exclusion
f = filter_builder.character_constraint("jack", True).get_character_filter()
self.assertTrue(f(el1))
self.assertTrue(f(el2))
self.assertTrue(f(el3))
filter_builder = FilterBuilder()
f = filter_builder.characters_as_disjunctive(True).character_constraint("john", True) \
.character_constraint("jack", True).get_character_filter()
self.assertFalse(f(el1))
self.assertTrue(f(el2))
self.assertTrue(f(el3))
def test_single_element_filter(self):
# Verify the effectiveness of single-valued matchers
id1, filename1, author1, universe1 = uuid4(), URI("01.png"), "bdhnd", "fotwf"
id2, filename2, author2, universe2 = uuid4(), URI("02.png"), "shndl", None
id3, filename3, author3, universe3 = uuid4(), URI("03.png"), "okn", "ph"
el1 = ImageMetadata(id1, filename1, author1, universe1, None, None)
el2 = ImageMetadata(id2, filename2, author2, universe2, None, None)
el3 = ImageMetadata(id3, filename3, author3, universe3, None, None)
filter_builder = FilterBuilder()
# Test constraints satisfied
filter_builder.filename_constraint(filename1.path.name) \
.filename_constraint(filename2.path.name) \
.filename_constraint(filename3.path.name)
filename_filter = filter_builder.get_filename_filter()
self.assertTrue(filename_filter(el1))
self.assertTrue(filename_filter(el2))
self.assertTrue(filename_filter(el3))
# Test implicit exclusion
filter_builder.author_constraint(author1)
author_filter = filter_builder.get_author_filter()
self.assertTrue(author_filter(el1))
self.assertFalse(author_filter(el2))
self.assertFalse(author_filter(el3))
# Test explicit exclusion
filter_builder.id_constraint(str(id2), True)
id_filter = filter_builder.get_id_filter()
self.assertTrue(id_filter(el1))
self.assertFalse(id_filter(el2))
self.assertTrue(id_filter(el3))
def test_empty_filter_single(self):
# Check the effects of the absence of constraints on single-valued matchers
el1 = ImageMetadata(uuid4(), URI('xxx'), None, None, None, None)
el2 = ImageMetadata(uuid4(), URI('kkk'), None, None, None, None)
empty_filter = FilterBuilder().get_id_filter()
self.assertTrue(empty_filter(el1))
self.assertTrue(empty_filter(el2))
def test_none_match_collective(self):
# Check the effects of None constraints on multi-valued matchers
el1 = ImageMetadata(uuid4(), URI('aaa'), None, None, None, None)
el2 = ImageMetadata(uuid4(), URI('yyy'), None, None, None, ["fta"])
none_filter = FilterBuilder().tag_constraint(None).get_tag_filter()
self.assertTrue(none_filter(el1))
self.assertFalse(none_filter(el2))
def test_empty_filter_collective(self):
# Check the effects of the absence of constraints on multi-valued matchers
el1 = ImageMetadata(uuid4(), URI('xxx'), None, None, None, ["nl", "ll"])
el2 = ImageMetadata(uuid4(), URI('kkk'), None, None, None, None)
empty_filter = FilterBuilder().get_tag_filter()
self.assertTrue(empty_filter(el1))
self.assertTrue(empty_filter(el2))
def test_none_match_single(self):
# Check the effects of None constraints on single-valued matchers
el1 = ImageMetadata(uuid4(), URI('fff'), None, 'u', None, None)
el2 = ImageMetadata(uuid4(), URI('zzz'), None, None, None, None)
none_filter = FilterBuilder().universe_constraint(None).get_universe_filter()
self.assertFalse(none_filter(el1))
self.assertTrue(none_filter(el2))
def test_without_optional_elements(self):
img_id = uuid.UUID('03012ba3-086c-4604-bd6a-aa3e1a78f389')
file = uri.URI("test.png")
self.assertEqual(
ImageMetadata(img_id, file, None, None, None, None),
xm.parse_xml(
xm.generate_xml(
ImageMetadata(img_id, file, None, None, None, None))))
def test_with_optional_elements(self):
img_id = uuid.UUID('03012ba3-086c-4604-bd6a-aa3e1a78f389')
file = uri.URI("test.png")
author = "John Doe"
universe = "Example"
characters = ["M", "Q"]
tags = ["a", "bee"]
self.assertEqual(
ImageMetadata(img_id, file, author, universe, characters, tags),
xm.parse_xml(
xm.generate_xml(
ImageMetadata(img_id, file, author, universe, characters,
tags))))
def _old_to_new_schema(img_path: Path, old_meta: ImageMetadata):
return ImageMetadata(img_id=old_meta.img_id,
file=URI(img_path),
author=old_meta.author,
universe=old_meta.universe,
characters=old_meta.characters,
tags=old_meta.tags)
def parse_xml(data: str) -> ImageMetadata:
"""Parse an XML containing image metadata.
:param data: a string containing valid image metadata
:return: an image metadata object"""
image_elem = ElTree.fromstring(data)
img_id = image_elem.get('id')
file = image_elem.get('file')
# If we were presented with a legacy XML not containing 'file', use the legacy name 'filename'
if file is None:
file = image_elem.get('filename')
author = image_elem.find("./author")
universe = image_elem.find("./universe")
characters = [
char.text for char in image_elem.findall("./characters/character")
]
tags = [tag.text for tag in image_elem.findall("./tags/tag")]
return ImageMetadata(
img_id=UUID(img_id),
file=URI(file),
author=author.text if author is not None else None,
universe=universe.text if universe is not None else None,
characters=characters if len(characters) != 0 else None,
tags=tags if len(tags) != 0 else None)
def test_generate_minimal(self):
img_id = uuid.UUID('03012ba3-086c-4604-bd6a-aa3e1a78f389')
file = uri.URI("test.png")
minimal_xml = xm.generate_xml(
ImageMetadata(img_id, file, None, None, None, None))
expected = '<image id="{i}" file="{f}" />'.format(i=img_id, f=file)
self.assertEqual(expected, minimal_xml)
def meta_extractor(v: View) -> ImageMetadata:
characters = v.get_characters()
characters = characters.split(', ') if characters is not None else None
tags = v.get_tags()
tags = tags.split(', ') if tags is not None else None
return ImageMetadata(v.image_id, URI(v._image_path), v.get_author(),
v.get_universe(), characters, tags)
def write(self) -> None:
"""
Persist the updated metadata.
:raise OSError: when the metadata file couldn't be opened
"""
meta_obj = ImageMetadata(self._id, URI(self._image_path), self.author,
self.universe, self.characters, self.tags)
write_meta(meta_obj, self._image_path)
def test_legacy_load(self):
image_uri = URI(self.test_path / "test.png")
with (self.test_path / "test.xml").open('w') as f:
f.write(
"<image id=\"97ed6183-73a0-46ea-b51d-0721b0fbd357\" filename=\"test.png\"></image>"
)
loaded = load_meta(Path(image_uri.path))
self.assertEqual(
ImageMetadata(UUID('97ed6183-73a0-46ea-b51d-0721b0fbd357'),
image_uri, None, None, None, None), loaded)
def test_load_actual(self):
image_uri = URI(self.test_path / "test.png")
with (self.test_path / "test.xml").open('w') as f:
f.write(
"<image id=\"97ed6183-73a0-46ea-b51d-0721b0fbd357\" file=\"" +
str(image_uri) + "\">" +
"<author>a</author><universe>u</universe>" +
"<characters><character>x</character><character>y</character></characters>"
+ "<tags><tag>f</tag><tag>a</tag></tags></image>")
self.assertEqual(
ImageMetadata(UUID('97ed6183-73a0-46ea-b51d-0721b0fbd357'),
image_uri, "a", "u", ["x", "y"], ["f", "a"]),
load_meta(Path(self.test_dir.name) / "test.png"))
def test_collective_conjunctive_filter(self):
# Verify the effectiveness of multi-valued matchers, when evaluated in conjunction
tags1 = ["y", "an", "hry"]
tags2 = ["an", "mb", "sty", "rp"]
tags3 = ["ll", "vnl"]
el1 = ImageMetadata(uuid4(), URI("a.png"), "ghi", None, None, tags1)
el2 = ImageMetadata(uuid4(), URI("b.png"), "nsh", None, None, tags2)
el3 = ImageMetadata(uuid4(), URI("c.png"), "ShT", None, None, tags3)
filter_builder = FilterBuilder()
# Test conjunctive filtering with inclusion
f = filter_builder.tag_constraint("an").get_tag_filter()
self.assertTrue(f(el1))
self.assertTrue(f(el2))
self.assertFalse(f(el3))
# Test conjunctive filtering with exclusion
f = filter_builder.tag_constraint("y", True).get_tag_filter()
self.assertFalse(f(el1))
self.assertTrue(f(el2))
self.assertFalse((f(el3)))
def test_filter(self):
filenames_meta = [("included.png", "included"),
("excluded.jpg", "excluded")]
for name, meta in filenames_meta:
img_path = (Path(self.test_dir.name) / name)
img_path.touch()
# Use the author field as the filtering target
write_meta(ImageMetadata(uuid4(), name, meta, None, None, None),
img_path)
specimen = Carousel(Path(self.test_dir.name),
[lambda meta: meta.author == "included"])
self.assertEqual([Path(self.test_dir.name) / "included.png"],
specimen._image_files)
def test_metadata_read(self):
# Write some metadata for one of the images
meta1 = ImageMetadata(
img_id=UUID('f32ed6ad-1162-4ea6-b243-1e6c91fb7eda'),
file=URI(self.test_path / '01.png'),
author="a",
universe="p",
characters=["x", "y"],
tags=["t", "f"])
write_meta(meta1, (self.test_path / '01.png'))
specimen = GtkView(self.test_path)
# Collect metadata from the specimen
results = dict()
for _ in range(0, 2):
specimen.load_next()
results[specimen.filename] = TestView.meta_extractor(specimen)
self.assertEqual(meta1, results["01.png"])
self.assertEqual(
ImageMetadata(results["02.png"].img_id,
URI(self.test_path / "02.png"), None, None, None,
None), results["02.png"])
def test_store(self):
image_uri = URI(self.test_path / "test.png")
write_meta(
ImageMetadata(UUID('97ed6183-73a0-46ea-b51d-0721b0fbd357'),
image_uri, "a", "u", ["x", "y"], ["f", "a"]),
Path(self.test_dir.name) / "test.png")
with (self.test_path / "test.xml") as f:
result = f.read_text()
self.assertEqual(
"<image id=\"97ed6183-73a0-46ea-b51d-0721b0fbd357\" file=\"" +
str(image_uri) + "\">" +
"<author>a</author><universe>u</universe>" +
"<characters><character>x</character><character>y</character></characters>"
+ "<tags><tag>f</tag><tag>a</tag></tags></image>", result)
def test_generate_single_element_and_subtree(self):
img_id = uuid.UUID('03012ba3-086c-4604-bd6a-aa3e1a78f389')
file = uri.URI("test.png")
universe = "unknown"
tags = ["a", "b"]
new_xml = xm.generate_xml(
ImageMetadata(img_id, file, None, universe, None, tags))
expected = ('<image id="{i}" file="{f}">' +
'<universe>{u}</universe>' +
'<tags><tag>{t1}</tag><tag>{t2}</tag></tags>' +
'</image>').format(i=img_id,
f=file,
u=universe,
t1=tags[0],
t2=tags[1])
self.assertEqual(expected, new_xml)
def test_metadata_update(self):
target_filename = '02.png'
specimen = GtkView(self.test_path)
# Scan until we find our target
specimen.load_next()
while specimen.filename != target_filename:
specimen.load_next()
# Verify that no metadata is present
self.assertIsNone(specimen.get_tags())
# Set metadata and check coherence
specimen.set_author(":DD")
specimen.set_universe("\tu")
specimen.set_characters("3, f,p\n")
specimen.set_tags("fa,s \vjo,\u200dl")
specimen.write()
self.assertEqual(
ImageMetadata(specimen.image_id,
URI(self.test_path / target_filename), ":DD", "u",
["3", "f", "p"], ["fa", "s jo", "l"]),
load_meta(self.test_path / target_filename))
def test_parse_legacy(self):
xml_string_legacy = "<image id=\"03012ba3-086c-4604-bd6a-aa3e1a78f389\" filename=\"tt.png\"></image>"
metadata = ImageMetadata(
uuid.UUID('03012ba3-086c-4604-bd6a-aa3e1a78f389'),
uri.URI("tt.png"), None, None, None, None)
self.assertEqual(metadata, xm.parse_xml(xml_string_legacy))
