def check(self, path):
# NOTE: this uses the package because of the way hathi packages are formatted
valid = True
errors = []
# Check if everything in access folder is found same in the preservation folder
missing_pres_files = self.check_for_missing_matching_preservation(
access_folder=path.directories["access"],
preservation_folder=path.directories["preservation"])
if missing_pres_files:
valid = False
new_error = error_message.ValidationError(
"The files [{}] were found in the access but not in the preservation folder"
.format(", ".join(
[os.path.basename(f) for f in missing_pres_files])),
group=path.identifier)
new_error.source = path.directories["access"]
errors.append(new_error)
missing_access_files = self.check_for_missing_matching_access(
access_folder=path.directories["access"],
preservation_folder=path.directories["preservation"])
if missing_access_files:
new_error = error_message.ValidationError(
"The files [{}] were found in the preservation folder but not in the access folder"
.format(", ".join(
[os.path.basename(f) for f in missing_access_files])),
group=path.identifier)
new_error.source = path.directories["preservation"]
errors.append(new_error)
return checkers.Results(self.checker_name(),
valid=valid,
errors=errors)
def check(self, path):
valid = True
errors = []
file_location = os.path.dirname(path)
basename, extension = os.path.splitext(os.path.basename(path))
if extension not in self.ignore_extension:
if extension not in self.valid_extensions:
valid = False
new_error = error_message.ValidationError(
"Invalid preservation file extension: \"{}\"".format(extension),
group=path)
new_error.source = path
errors.append(new_error)
# Check the image files have the full 8 digits
if extension == ".tif":
if "target" not in basename:
if PresNamingChecker.valid_naming_scheme.match(basename) is None:
valid = False
new_error = error_message.ValidationError(
"Does not match the valid preservation file naming pattern",
group=file_location.split(os.sep)[-1])
new_error.source = path
errors.append(new_error)
return checkers.Results(self.checker_name(), valid=valid, errors=errors)
def check(self, path: str):
"""
Make sure that all files included in this folder are tiff files
and contain nothing else
Args:
path: Path to the folder to check
Returns: list of errors
"""
required_files = set() # type: ignore
required_files = {"checksum.md5", "marc.xml", "meta.yml"}
valid_image_extensions = [".jp2"]
valid_text_extensions = [".txt", ".xml", ".yml"]
errors = []
valid = True
image_files = set()
text_files = set()
try:
missing = list(self.find_missing_by_number(path))
if missing:
valid = False
new_error = error_message.ValidationError(
"Expected files [{}] not found in access folder".format(", ".join(missing)),
group=path.split(os.sep)[-1])
new_error.source = path
errors.append(new_error)
except ValueError as e:
valid = False
new_error = error_message.ValidationError("Error trying to find missing files. Reason: {}".format(e),
group=path.split(os.sep)[-1])
new_error.source = path
errors.append(new_error)
# Sort the files into their own category
for root, dirs, files in os.walk(path):
for file_ in files:
# if the filename is the required files set, remove them
if file_ in required_files:
required_files.remove(file_)
basename, ext = os.path.splitext(file_)
if ext in valid_image_extensions:
image_files.add((root, file_))
elif ext in valid_text_extensions:
text_files.add((root, file_))
# If there are any files still in the required_files set are missing.
if required_files:
valid = False
new_error = error_message.ValidationError(
"Missing expected file(s), [{}]".format(", ".join(required_files)))
new_error.source = path
errors.append(new_error)
# errors.append("{} is missing {}".format(path, _file))
return checkers.Results(self.checker_name(), valid=valid, errors=errors)
def check(self, path):
valid = True
errors = []
for error in self.find_root_directory_errors(path):
valid = False
errors.append(error)
if valid:
for error in self.find_subdirectory_errors(path):
valid = False
errors.append(error)
return checkers.Results(self.checker_name(),
valid=valid,
errors=errors)
def check(self, path):
valid = True
errors = []
required_files = (
"target_l_001.tif",
"target_l_002.tif",
"target_r_001.tif",
"target_r_002.tif",
)
error_group = path.split(os.sep)[-1]
try:
missing = list(self.find_missing_by_number(path))
if missing:
valid = False
new_error = error_message.ValidationError(
"Expected files [{}] not found in preservation folder".
format(", ".join(missing)),
group=error_group)
new_error.source = path
errors.append(new_error)
except ValueError as e:
valid = False
new_error = error_message.ValidationError(
"Error trying to find missing files. Reason: {}".format(e),
group=error_group)
new_error.source = path
errors.append(new_error)
except FileNotFoundError as e:
valid = False
new_error = error_message.ValidationError(e, group=error_group)
new_error.source = path
errors.append(new_error)
# return checkers.Results(self.checker_name(), valid=valid, errors=errors)
# Find missing required_files
missing = list(
self.find_missing_required_files(path=path,
expected_files=required_files))
if missing:
valid = False
new_error = error_message.ValidationError(
"Missing expected file(s), [{}]".format(", ".join(missing)),
group=error_group)
new_error.source = path
errors.append(new_error)
return checkers.Results(self.checker_name(),
valid=valid,
errors=errors)
def check(self, path):
valid = True
errors = []
file_location = os.path.dirname(path)
group_name = file_location.split(os.sep)[-1]
basename, extension = os.path.splitext(os.path.basename(path))
if extension in self.extensions_to_check:
if self.valid_naming_scheme.match(basename) is None:
valid = False
new_error = error_message.ValidationError(
"Does not match the valid file pattern for preservation files",
group=group_name)
new_error.source = path
errors.append(new_error)
return checkers.Results(self.checker_name(), valid=valid, errors=errors)
def check(self, path):
valid = True
errors = []
file_location = os.path.dirname(path)
basename, extension = os.path.splitext(os.path.basename(path))
if extension not in self.ignore_extension:
if extension not in self.valid_extensions:
valid = False
new_error = error_message.ValidationError("Invalid file type",
group=path.split(
os.sep)[-1])
new_error.source = path
errors.append(new_error)
# Check the image files have the full 8 digits
if self.valid_naming_scheme.match(basename) is None:
valid = False
new_error = error_message.ValidationError(
"Does not match the valid file pattern for preservation files",
group=file_location.split(os.sep)[-1])
new_error.source = path
errors.append(new_error)
#
# # The only xml file should be marc.xml
# if extension == ".xml":
# if basename != "marc":
# valid = False
# errors.append(
# "\"{}\" does not match the valid file pattern for preservation files".format(basename))
#
# # The only yml file should be meta.yml
# if extension == ".yml":
# if basename != "meta":
# valid = False
# errors.append(
# "\"{}\" does not match the valid file result_type pattern for preservation files".format(basename))
return checkers.Results(self.checker_name(),
valid=valid,
errors=errors)
def check(self, path):
valid = True
errors = []
return checkers.Results(self.checker_name(), valid=valid, errors=errors)
def check(self, path):
# NOTE: this uses the package because of the way hathi packages are formatted
valid = True
errors = []
return checkers.Results(self.checker_name(), valid=valid, errors=errors)