def transformed_to_dl_compound(
self, sample_cataloged_collection, monkeypatch):
root, files = sample_cataloged_collection
factory = packager.PackageFactory(noneas.CatalogedNonEAS())
digital_library_format = packager.PackageFactory(
digital_library_compound.DigitalLibraryCompound())
output_path = os.path.join('some', 'folder')
transform = Mock()
def mock_transform(_, source, destination):
transform(source=source, destination=destination)
with monkeypatch.context() as mp:
mp.setattr(
packager.transformations.Transformers,
"transform", mock_transform
)
monkeypatch.setattr(
uiucprescon.packager.transformations.CopyFile,
"transform",
transform
)
monkeypatch.setattr(
uiucprescon.packager.transformations.ConvertJp2Standard,
"transform",
transform
)
for package in list(factory.locate_packages(root)):
digital_library_format.transform(package, output_path)
return output_path, transform
def transformed_to_dl_compound(
self, sample_collection, monkeypatch
):
root, files = sample_collection
factory = packager.PackageFactory(noneas.ArchivalNonEAS())
digital_library_format = packager.PackageFactory(
digital_library_compound.DigitalLibraryCompound())
import shutil
kdu_compress_cli2 = Mock()
output_path = os.path.join('some', 'folder')
transform = Mock()
with monkeypatch.context() as mp:
mp.setattr(
packager.transformations.Transformers, "transform", transform
)
mp.setattr(
uiucprescon.packager.transformations.CopyFile,
"transform",
transform
)
mp.setattr(
uiucprescon.packager.transformations.ConvertJp2Standard,
"transform",
transform
)
mp.setattr(shutil, "copyfile", Mock())
mp.setattr(shutil, "copy", Mock())
import pykdu_compress
mp.setattr(pykdu_compress, "kdu_compress_cli2", kdu_compress_cli2)
for package in list(factory.locate_packages(root)):
digital_library_format.transform(package, output_path)
return output_path, transform
def test_transform_to_hathi(self,
sample_collection_path,
expected_source,
expected_destination,
monkeypatch):
factory = packager.PackageFactory(eas.Eas())
packages = factory.locate_packages(sample_collection_path)
destination_type = packager.PackageFactory(HathiJp2())
output = "out"
def spec(source, destination): pass
transform = Mock(spec=spec)
monkeypatch.setattr(
packager.transformations.Transformers,
"transform",
transform
)
monkeypatch.setattr(
packager.packages.hathi_jp2_package.pathlib.Path,
"mkdir",
Mock()
)
for p in packages:
destination_type.transform(p, output)
assert transform.called is True
transform.assert_has_calls(
[
call(source=expected_source, destination=expected_destination)
]
)
def test_capture_one_tiff_to_hathi_tiff(capture_one_fixture):
source = os.path.join(capture_one_fixture, CAPTURE_ONE_BATCH_NAME)
dest = os.path.join(capture_one_fixture, DESTINATION_NAME)
capture_one_packages_factory = \
packager.PackageFactory(CaptureOnePackage(delimiter="_"))
# find all Capture One organized packages
capture_one_packages = \
list(capture_one_packages_factory.locate_packages(path=source))
# There should be 2 packages in this sample batch
assert len(capture_one_packages) == 2
hathi_tiff_package_factory = \
packager.PackageFactory(packager.packages.HathiTiff())
for cap_one_package in capture_one_packages:
hathi_tiff_package_factory.transform(cap_one_package, dest=dest)
# This should result in the following files
#
# some_root/000001/00000001.tif
# some_root/000001/00000002.tif
# some_root/000001/00000003.tif
assert os.path.exists(os.path.join(dest, "000001", "00000001.tif"))
assert os.path.exists(os.path.join(dest, "000001", "00000002.tif"))
assert os.path.exists(os.path.join(dest, "000001", "00000003.tif"))
# some_root/000002/00000001.tif
# some_root/000002/00000002.tif
assert os.path.exists(os.path.join(dest, "000002", "00000001.tif"))
assert os.path.exists(os.path.join(dest, "000002", "00000002.tif"))
def test_read_only_transform(capture_one_sample_package):
capture_one_packager = packager.PackageFactory(
packager.packages.CaptureOnePackage())
capture_one_packages = capture_one_packager.locate_packages(
capture_one_sample_package)
hathi_limited_view_packager = packager.PackageFactory(
packager.packages.HathiLimitedView())
with pytest.raises(NotImplementedError):
hathi_limited_view_packager.transform(capture_one_packages, dest=".")
def discover_task_metadata(self,
initial_results: List[Any],
additional_data: Dict[str, Any],
**user_args: str
) -> List[Dict[str, Any]]:
"""Loot at user settings and discover any data needed to build a task.
Args:
initial_results:
additional_data:
**user_args:
Returns:
Returns a list of data to create a job with
"""
jobs: List[Dict[str, Union[str, Package]]] = []
source_input = user_args["Input"]
dest = user_args["Output"]
package_factory = packager.PackageFactory(
packager.packages.CaptureOnePackage(delimiter="-"))
for package in package_factory.locate_packages(source_input):
new_job: Dict[str, Union[str, Package]] = {
"package": package,
"output": dest,
"source_path": source_input
}
jobs.append(new_job)
return jobs
def package_objects(source_path, source_package_type):
source_pkg = eval(f"packager.packages.{source_package_type}")
source = os.path.join(source_path, sample_packages[source_package_type][0])
packages_factory = packager.PackageFactory(source_pkg())
packages = list(packages_factory.locate_packages(path=source))
return packages
def test_cataloged_collection_transform(
self,
eas_collection,
source_file, package_type, expected_out,
monkeypatch
):
transform = Mock(spec=lambda source, destination: None)
monkeypatch.setattr(
packager.transformations.Transformers,
"transform",
transform
)
output_dir = "output"
factory = packager.PackageFactory(eas.Eas())
for p in factory.locate_packages(eas_collection(source_file)):
packager.PackageFactory(package_type()).transform(p, output_dir)
def test_capture_one_dashes(capture_one_batch_with_dashes):
batch_dir, source_files = capture_one_batch_with_dashes
capture_one_packages_factory = packager.PackageFactory(
packager.packages.CaptureOnePackage(delimiter="-"))
res = next(capture_one_packages_factory.locate_packages(batch_dir))
assert len(res) == len(source_files)
def test_capture_one_underscore(capture_one_batch_with_underscores):
batch_dir, source_files = capture_one_batch_with_underscores
capture_one_packages_factory = packager.PackageFactory(
packager.packages.CaptureOnePackage()
)
res = next(capture_one_packages_factory.locate_packages(batch_dir))
assert len(res) == len(source_files)
def capture_one_session_w_ds_store(capture_one_sample_package):
source_dir = os.path.join(capture_one_sample_package)
with open(os.path.join(source_dir, ".DS_Store"), "w") as wf:
pass
capture_one_packages_factory = \
packager.PackageFactory(packager.packages.CaptureOnePackage())
# find all Capture One organized packages
return list(capture_one_packages_factory.locate_packages(path=source_dir))
def package_objects(source_path, package_type):
pkg_factory_type = eval(f"packager.packages.{package_type}")
source = os.path.join(source_path, sample_packages[package_type][0])
capture_one_packages_factory = packager.PackageFactory(pkg_factory_type())
# find all Capture One organized packages
packages = list(capture_one_packages_factory.locate_packages(path=source))
return packages
def __init__(
self,
src: packager.package.collection.Package,
dst: str
) -> None:
super().__init__()
self.src = src
self.dst = dst
self.output_packager = packager.PackageFactory(
packager.packages.DigitalLibraryCompound())
def test_transform_into_hathi(capture_one_batch_with_dashes, tmpdir):
batch_dir, source_files = capture_one_batch_with_dashes
source_type = packager.PackageFactory(
packager.packages.CaptureOnePackage(delimiter="-")
)
packages = source_type.locate_packages(batch_dir)
destination_type = packager.PackageFactory(packager.packages.HathiTiff())
output = tmpdir / "output"
output.ensure_dir()
for package in packages:
destination_type.transform(package, dest=output.strpath)
assert (output / "99423682912205899").exists()
for expected_file in [f"{str(x).zfill(8)}.tif" for x in range(20)]:
assert (output / "99423682912205899" / expected_file).exists()
output.remove()
def work(self) -> bool:
self.log("Locating packages in {}".format(self._root))
package_factory = packager.PackageFactory(
packager.packages.CaptureOnePackage())
packages = list(package_factory.locate_packages(self._root))
self.set_results(packages)
return True
def test_cataloged_collection_transform(
self,
cataloged_collection,
source_file, package_type, expected_out,
monkeypatch
):
transform = Mock(spec=lambda source, destination: None)
monkeypatch.setattr(
uiucprescon.packager.transformations.Transformers,
"transform",
transform
)
transform2 = Mock(spec=lambda source, destination, logger: None)
monkeypatch.setattr(
uiucprescon.packager.transformations.CopyFile,
"transform",
transform2
)
monkeypatch.setattr(
uiucprescon.packager.transformations.ConvertJp2Standard,
"transform",
transform2
)
output_dir = "output"
factory = packager.PackageFactory(noneas.CatalogedNonEAS())
for p in factory.locate_packages(cataloged_collection(source_file)):
packager.PackageFactory(package_type()).transform(p, output_dir)
if package_type == hathi_jp2_package.HathiJp2:
transform.assert_any_call(
ANY, os.path.join(output_dir, expected_out)
)
elif package_type == digital_library_compound.DigitalLibraryCompound:
transform2.assert_any_call(
ANY, os.path.join(output_dir, expected_out), ANY
)
else:
assert False, f"testing '{package_type}' not supported"
def test_capture_one_tiff_package_size(capture_one_fixture):
source = os.path.join(capture_one_fixture, CAPTURE_ONE_BATCH_NAME)
capture_one_packages_factory = \
packager.PackageFactory(packager.packages.CaptureOnePackage())
# find all Capture One organized packages
capture_one_packages = \
list(capture_one_packages_factory.locate_packages(path=source))
# There should be 2 packages in this sample batch
assert len(capture_one_packages) == 2
def work(self):
my_logger = logging.getLogger(packager.__name__)
my_logger.setLevel(logging.INFO)
with self.log_config(my_logger):
self.log(f"Converting {self.packaging_id} from {self.source_path} "
f"to a {self.package_format} package at "
f"{self.new_package_root}")
package_factory = packager.PackageFactory(
PackageConverter.package_formats[self.package_format])
package_factory.transform(self.existing_package,
dest=self.new_package_root)
return True
def work(self) -> bool:
my_logger = logging.getLogger(packager.__name__)
my_logger.setLevel(logging.INFO)
with self.log_config(my_logger):
self.log(f"Converting {self.packaging_id} from "
f"{self.source_path} to a Hathi Trust Tiff "
f"package at {self.new_package_root}")
package_factory = packager.PackageFactory(
packager.packages.HathiTiff())
package_factory.transform(self.existing_package,
dest=self.new_package_root)
return True
def test_convert(hathi_limited_view_sample_packages, monkeypatch):
import pathlib
def kdu_compress_cli2(infile: str,
outfile: str,
in_args=None,
out_args=None):
pathlib.Path(outfile).touch()
def kdu_expand_cli(infile: str, outfile: str, in_args=None, out_args=None):
pathlib.Path(outfile).touch()
monkeypatch.setattr(pykdu_compress, "kdu_compress_cli2", kdu_compress_cli2)
monkeypatch.setattr(pykdu_compress, "kdu_expand_cli", kdu_expand_cli)
digital_library_compound_builder = packager.PackageFactory(
packager.packages.DigitalLibraryCompound())
with tempfile.TemporaryDirectory() as tmp_dir:
for package in hathi_limited_view_sample_packages:
try:
digital_library_compound_builder.transform(package,
dest=tmp_dir)
except errors.ZipFileException as e:
print(f"{e.src_zip_file} had a problem", file=sys.stderr)
if len(e.problem_files) > 0:
print(f"Problems with {','.join(e.problem_files)}",
file=sys.stderr)
problem_file = zipfile.ZipFile(e.src_zip_file)
print(problem_file.namelist(), file=sys.stderr)
raise
assert len(list(os.scandir(tmp_dir))) == 1
for i, new_package in enumerate(
digital_library_compound_builder.locate_packages(tmp_dir)):
assert new_package.metadata[Metadata.ID] == \
hathi_limited_view_sample_packages[i].metadata[Metadata.ID]
sample_item = new_package.items[0]
access = sample_item.instantiations[InstantiationTypes.ACCESS]
access_files = list(access.get_files())
assert len(access_files) > 0
pres = sample_item.instantiations[InstantiationTypes.PRESERVATION]
pres_files = list(pres.get_files())
assert len(pres_files) > 0
assert new_package.metadata[Metadata.PATH] == tmp_dir
def test_capture_one_collection_transform(
self,
capture_one_collection,
source_file, package_type, expected_out,
monkeypatch
):
transform = Mock(spec=lambda source, destination: None)
monkeypatch.setattr(
uiucprescon.packager.transformations.Transformers,
"transform",
transform
)
output_dir = "output"
capture_one_factory = packager.PackageFactory(
packager.packages.CaptureOnePackage(delimiter='-')
)
for p in capture_one_factory.locate_packages(
capture_one_collection(source_file)
):
packager.PackageFactory(package_type()).transform(p, output_dir)
#
transform.assert_any_call(ANY, os.path.join(output_dir, expected_out))
def discover_task_metadata(
self,
initial_results: List[Any],
additional_data, **user_args: str
) -> List[dict]:
hathi_limited_view_packager = packager.PackageFactory(
packager.packages.HathiLimitedView())
return [{
"package": package,
"destination": user_args['Output']
} for package in hathi_limited_view_packager.locate_packages(
user_args['Input'])]
def work(self):
my_logger = logging.getLogger(packager.__name__)
my_logger.setLevel(logging.INFO)
with self.log_config(my_logger):
self.log(
f"Converting {self.packaging_id} from {self.source_path} "
f"to a Hathi Trust Tiff package at {self.new_package_root}")
package_factory = packager.PackageFactory(
packager.packages.DigitalLibraryCompound())
package_factory.transform(self.existing_package,
dest=self.new_package_root)
return True
def work(self) -> bool:
package_factory = packager.PackageFactory(
packager.packages.HathiTiff())
package_factory.transform(self._package, self._destination)
self.log("Transformed CaptureOne package {} to a HathiTiff package "
"in {}".format(self._bib_id, self._destination))
self.set_results({
"bib_id":
self._bib_id,
"location":
os.path.join(self._destination, self._bib_id)
})
return True
def hathi_tiff_package_w_sidecar_text(hathi_tiff_sample_package):
package_one_path = os.path.join(hathi_tiff_sample_package, "000001")
for i in range(3):
with open(os.path.join(package_one_path, f"0000000{i+1}.txt"), "w"):
pass
package_two_path = os.path.join(hathi_tiff_sample_package, "000002")
for i in range(2):
with open(os.path.join(package_two_path, f"0000000{i+1}.txt"), "w"):
pass
package_factory = packager.PackageFactory(packager.packages.HathiTiff())
packages = list(package_factory.locate_packages(hathi_tiff_sample_package))
return packages
def archival_full_name_transformed_to_ht_trust(
self, sample_collection_longer, monkeypatch):
root, files = sample_collection_longer
factory = packager.PackageFactory(noneas.ArchivalNonEAS())
hathi_jp2_format = packager.PackageFactory(
hathi_jp2_package.HathiJp2()
)
output_path = os.path.join('some', 'folder')
transform = Mock()
def mock_transform(_, source, destination):
transform(source=source, destination=destination)
with monkeypatch.context() as mp:
mp.setattr(
packager.transformations.Transformers,
"transform", mock_transform
)
for package in factory.locate_packages(root):
hathi_jp2_format.transform(package, output_path)
return output_path, transform
def transformed_to_ht_trust(
self, sample_cataloged_collection, monkeypatch, tmpdir
):
transform = Mock()
def mock_transform(_, source, destination):
transform(source=source, destination=destination)
root, files = sample_cataloged_collection
factory = packager.PackageFactory(noneas.CatalogedNonEAS())
digital_library_format = packager.PackageFactory(
hathi_jp2_package.HathiJp2())
output_path = os.path.join('some', 'folder')
with monkeypatch.context() as mp:
mp.setattr(
packager.transformations.Transformers,
"transform", mock_transform
)
for package in list(factory.locate_packages(root)):
digital_library_format.transform(package, output_path)
return output_path, transform
def discover_task_metadata(self, initial_results: List[Any],
additional_data, **user_args) -> List[dict]:
hathi_limited_view_packager = packager.PackageFactory(
packager.packages.HathiLimitedView())
new_tasks = []
for p in hathi_limited_view_packager.locate_packages(
user_args['Input']):
new_tasks.append({
"package": p,
"destination": user_args['Output']
})
return new_tasks
def discover_task_metadata(self, initial_results: List[Any],
additional_data, **user_args) -> List[dict]:
jobs = []
source_input = user_args["Input"]
dest = user_args["Output"]
package_factory = packager.PackageFactory(
packager.packages.CaptureOnePackage())
for package in package_factory.locate_packages(source_input):
jobs.append({
"package": package,
"output": dest,
"source_path": source_input
})
return jobs
def test_capture_one_tiff_package_plus(capture_one_fixture_plus):
capture_one_packages_factory = \
packager.PackageFactory(
packager.packages.CaptureOnePackage(delimiter='+')
)
# find all Capture One organized packages
capture_one_packages = \
list(
capture_one_packages_factory.locate_packages(
path=capture_one_fixture_plus
)
)
# There should be 2 packages in this sample batch
assert len(capture_one_packages) == 2