def item(self):
item = collection.Item()
item.component_metadata[packager.Metadata.ID] = '99127822912205899'
item.component_metadata[packager.Metadata.ITEM_NAME] = '001'
item.component_metadata[packager.Metadata.PATH] = 'somepath'
access_file = \
collection.Instantiation(
parent=item,
category=packager.InstantiationTypes.ACCESS,
files=["99127822912205899-001.tif"]
)
access_file.component_metadata[packager.Metadata.PATH] = \
os.path.join("somepath", "access")
preservation_file = \
collection.Instantiation(
parent=item,
category=packager.InstantiationTypes.PRESERVATION,
files=["99127822912205899-001.tif"]
)
preservation_file.component_metadata[packager.Metadata.PATH] = \
os.path.join("somepath", "preservation")
supplementary_data_files = collection.Instantiation(
parent=item,
category=packager.InstantiationTypes.SUPPLEMENTARY,
files=[
"99127822912205899-001a.txt",
"99127822912205899-001b.txt",
])
supplementary_data_files.component_metadata[packager.Metadata.PATH] = \
"somepath"
return item
def test_transform_access_error_on_multiple_access(self):
compound_item = collection.Item()
compound_item.component_metadata[collection.Metadata.ID] = "123"
compound_item.component_metadata[collection.Metadata.PATH] = "source"
compound_item.component_metadata[collection.Metadata.ITEM_NAME] \
= '00000001'
collection.Instantiation(category=collection.InstantiationTypes.ACCESS,
parent=compound_item,
files=[
os.path.join("123", "access",
"123-00000001.jp2"),
os.path.join("123", "access",
"123-00000001a.jp2")
])
collection.Instantiation(
category=collection.InstantiationTypes.PRESERVATION,
parent=compound_item,
files=[
os.path.join("123", "preservation", "123-00000001.tif"),
os.path.join("123", "preservation", "123-00000001a.tif"),
])
strategy = packages.hathi_jp2_package.CopyStrategy()
strategy.convert = Mock()
with pytest.raises(AssertionError):
strategy.transform_access_file(compound_item, dest="out")
def test_set_title_page(self, delegate):
combo_box = QtWidgets.QComboBox()
files = [
"file1.jp2",
"file2.jp2",
"file3.jp2",
]
for file_name in files:
combo_box.addItem(file_name)
combo_box.setCurrentText("file2.jp2")
model = Mock()
object_record = collection.PackageObject()
item = collection.Item(object_record)
instance = collection.Instantiation(parent=item, files=files)
def get_data(index, role):
if role == QtCore.Qt.UserRole:
return object_record
mock_index = MagicMock()
model.data = get_data
delegate.setModelData(combo_box, model, mock_index)
assert \
object_record.metadata[collection.Metadata.TITLE_PAGE] == \
"file2.jp2"
def test_builder2_build_instance(capture_one_batch_with_dashes):
batch_dir, source_files = capture_one_batch_with_dashes
builder = capture_one_package.CaptureOneBuilder()
builder.splitter = capture_one_package.dash_splitter
sample_item = collection.Item()
sample_item.component_metadata[Metadata.ID] = "00001"
builder.build_instance(
parent=sample_item, path=batch_dir, filename=source_files[0]
)
assert len(sample_item) == 1
def get_data(role):
if role == QtCore.Qt.UserRole:
object_record = collection.PackageObject()
item = collection.Item(object_record)
instance = collection.Instantiation(parent=item,
files=[
"file1.jp2",
"file2.jp2",
"file3.jp2",
])
return object_record
def get_data(role):
if role == QtCore.Qt.UserRole:
object_record = collection.PackageObject()
object_record.component_metadata[
collection.Metadata.TITLE_PAGE] = "file2.jp2"
item = collection.Item(object_record)
instance = collection.Instantiation(parent=item,
files=[
"file1.jp2",
"file2.jp2",
"file3.jp2",
])
return object_record
def capture_one_item():
capture_one_tiff = collection.Item()
capture_one_tiff.component_metadata[collection.Metadata.ID] = "123"
capture_one_tiff.component_metadata[collection.Metadata.PATH] = "source"
capture_one_tiff.component_metadata[
collection.Metadata.ITEM_NAME] = '00000001'
collection.Instantiation(
category=collection.InstantiationTypes.PRESERVATION,
parent=capture_one_tiff,
files=[os.path.join("123", "123_00000001.tif")])
return capture_one_tiff
def compound_item():
new_item = collection.Item()
new_item.component_metadata[collection.Metadata.ID] = "123"
new_item.component_metadata[collection.Metadata.PATH] = "source"
new_item.component_metadata[collection.Metadata.ITEM_NAME] = '00000001'
collection.Instantiation(
category=collection.InstantiationTypes.ACCESS,
parent=new_item,
files=[os.path.join("123", "access", "123-00000001.jp2")])
collection.Instantiation(
category=collection.InstantiationTypes.PRESERVATION,
parent=new_item,
files=[os.path.join("123", "preservation", "123-00000001.tif")])
return new_item
def test_zip_error(monkeypatch):
item = collection.Item()
item.component_metadata[Metadata.PATH] = "somepath.zip"
instance = collection.Instantiation(parent=item)
instance._files = [
"somefile.txt"
]
from zipfile import ZipFile
import io
with monkeypatch.context() as mp:
mp.setattr(ZipFile, "extract", Mock(side_effect=KeyError))
mp.setattr(ZipFile, "_RealGetContents", Mock())
mp.setattr(io, "open", MagicMock())
with pytest.raises(errors.ZipFileException) as error:
next(instance.get_files())
assert error.value.src_zip_file == item.metadata[Metadata.PATH] and \
"somefile.txt" in error.value.problem_files[0]
def build_package(self, parent, path: str, *args, **kwargs) -> None:
"""Build a capture one style package object."""
group_id = parent.metadata[Metadata.ID]
non_system_files = self.get_non_system_files(path)
def filter_by_group(candidate_file: "os.DirEntry[str]") -> bool:
parts = self.identify_file_name_parts(candidate_file.name)
return parts is not None and parts['group'] == group_id
for file_ in filter(filter_by_group, non_system_files):
file_name_parts = self.identify_file_name_parts(file_.name)
if file_name_parts is None:
raise ValueError(
f"File does not match expected naming pattern {file_.path}"
)
if file_name_parts['extension'].lower() != ".tif":
continue
item_part = file_name_parts['part']
new_item = collection.Item(parent=parent)
new_item.component_metadata[Metadata.ITEM_NAME] = item_part
self.build_instance(new_item, path, item_part)
def build_object(self,
parent: collection.Package,
group_id: str,
path: str) -> None:
"""Build a new object.
Args:
parent:
group_id:
path:
"""
new_object = collection.PackageObject(parent=parent)
new_object.component_metadata[Metadata.ID] = group_id
new_object.component_metadata[Metadata.PATH] = path
access_path = os.path.join(path, "access")
for directory_item in filter(
lambda item: self.is_eas_file(pathlib.Path(item.path)),
self.locate_files_access(access_path)
):
regex_result = EASBuilder.grouper_regex.match(directory_item.name)
if regex_result is None:
raise ValueError("Unknown pattern")
groups = regex_result.groupdict()
if groups['group'] != group_id:
continue
new_item = collection.Item(parent=new_object)
new_item.component_metadata[Metadata.ITEM_NAME] = \
regex_result['part']
self.build_instance(
new_item,
path=os.path.dirname(directory_item.path),
filename=directory_item.name
)
def test_transform_access_error_on_multiple_access(self):
capture_one_tiff = collection.Item()
capture_one_tiff.component_metadata[collection.Metadata.ID] = "123"
capture_one_tiff.component_metadata[collection.Metadata.PATH] \
= "source"
capture_one_tiff.component_metadata[
collection.Metadata.ITEM_NAME] = '00000001'
collection.Instantiation(
category=collection.InstantiationTypes.PRESERVATION,
parent=capture_one_tiff,
files=[
os.path.join("123", "123_00000001.tif"),
os.path.join("123", "123_00000001b.tif")
])
strategy = packages.hathi_jp2_package.ConvertStrategy(
instance_source=collection.InstantiationTypes.PRESERVATION)
strategy.convert = Mock()
with pytest.raises(AssertionError):
strategy.transform_access_file(capture_one_tiff, dest="out")
def digital_library_object(self):
new_object = collection.PackageObject()
new_object.component_metadata[collection.Metadata.ID] = "123"
new_object.component_metadata[collection.Metadata.PACKAGE_TYPE] = \
common.PackageTypes.DIGITAL_LIBRARY_COMPOUND
new_item = collection.Item(parent=new_object)
new_item.component_metadata[collection.Metadata.ID] = "123"
new_item.component_metadata[collection.Metadata.PATH] = "source"
new_item.component_metadata[collection.Metadata.ITEM_NAME] = '00000001'
collection.Instantiation(
category=collection.InstantiationTypes.ACCESS,
parent=new_item,
files=[os.path.join("123", "access", "123-00000001.jp2")])
collection.Instantiation(
category=collection.InstantiationTypes.PRESERVATION,
parent=new_item,
files=[os.path.join("123", "preservation", "123-00000001.tif")])
return new_object