def main():
parser = _make_parser()
args = parser.parse_args()
_configure_logging(args)
excel_paths = []
if args.excel:
excel_paths.append(args.excel)
if args.directory:
directory_path = os.path.abspath(args.directory)
for path in glob.glob("{}*.xls*".format(directory_path)):
path = os.path.join(directory_path, path)
excel_paths.append(path)
if args.output:
output_path = os.path.abspath(args.output)
for excel_path in excel_paths:
excel = ami_excel(excel_path)
print(excel_path)
excel.edit_sheet.add_PMDataToEM(excel.pres_sheet.sheet_values)
excel.edit_sheet.convert_amiExcelToJSON(output_path)
excel.pres_sheet.convert_amiExcelToJSON(output_path)
def main():
parser = _make_parser()
args = parser.parse_args()
_configure_logging(args)
excel_paths = []
if args.excel:
excel_paths.append(args.excel)
if args.directory:
directory_path = os.path.abspath(args.directory)
for path in glob.glob("{}*.xls*".format(directory_path)):
path = os.path.join(directory_path, path)
excel_paths.append(path)
if args.output:
output_path = os.path.abspath(args.output)
for excel_path in excel_paths:
csv_name = os.path.splitext(os.path.split(excel_path)[1])[0]
output_path = os.path.join(output_path, csv_name + '.csv')
excel = ami_excel(excel_path)
print(excel_path)
print(output_path)
excel.pres_sheet.convert_amiExcelToCSV(output_path)
def add_json_from_excel(self):
self.excel_metadata = [
filename for filename in self.data_files
if os.path.splitext(filename)[1] == ".xlsx"
]
for filename in self.excel_metadata:
excel = ami_excel(os.path.join(self.path, filename))
if excel.edit_sheet:
em_path = os.path.join(self.path, "data/EditMasters")
# TODO where do i error when files don't match
try:
excel.edit_sheet.add_PMDataToEM(
excel.pres_sheet.sheet_values)
except:
LOGGER.error(
"EM's and PM's do not have 1-1 correspondence")
else:
em_filepaths = [
x + ".json" for x in self.media_filepaths
if em_path in x
]
excel.edit_sheet.convert_amiExcelToJSON(
em_path, filepaths=em_filepaths)
pm_path = os.path.join(self.path, "data/PreservationMasters")
pm_filepaths = [
x + ".json" for x in self.media_filepaths if pm_path in x
]
excel.pres_sheet.convert_amiExcelToJSON(pm_path,
filepaths=pm_filepaths)
def set_metadata_excel(self):
self.metadata_files = [
filename for filename in self.data_files
if os.path.splitext(filename)[1] == ".xlsx"
]
self.media_files_md = []
for filename in self.metadata_files:
excel = ami_excel(os.path.join(self.path, filename))
# collect list of filenames in metadata
if excel.pres_sheet:
paths = excel.pres_sheet.sheet_values[
"asset.referenceFilename"].tolist()
self.media_files_md.extend(paths)
if excel.edit_sheet:
if "asset.referenceFilename" in excel.edit_sheet.sheet_values.columns:
paths = excel.edit_sheet.sheet_values[
"asset.referenceFilename"].tolist()
self.media_files_md.extend(paths)
self.media_files_md = set(self.media_files_md)
return
def main():
parser = _make_parser()
args = parser.parse_args()
_configure_logging(args)
excel_paths = []
if args.excel:
excel_paths.append(args.excel)
if args.directory:
directory_path = os.path.abspath(args.directory)
for path in glob.glob("{}*.xls*".format(directory_path)):
path = os.path.join(directory_path, path)
excel_paths.append(path)
if args.output:
output_path = os.path.abspath(args.output)
for excel_path in excel_paths:
excel = ami_excel(excel_path)
print(excel_path)
excel.edit_sheet.add_PMDataToEM(excel.pres_sheet.sheet_values)
excel.edit_sheet.convert_amiExcelToJSON(output_path)
excel.pres_sheet.convert_amiExcelToJSON(output_path)
def main():
parser = _make_parser()
args = parser.parse_args()
_configure_logging(args)
excel_paths = []
if args.excel:
excel_paths.append(args.excel)
if args.directory:
directory_path = os.path.abspath(args.directory)
for path in glob.glob("{}*.xls*".format(directory_path)):
path = os.path.join(directory_path, path)
excel_paths.append(path)
if args.output:
output_path = os.path.abspath(args.output)
for excel_path in excel_paths:
csv_name = os.path.splitext(os.path.split(excel_path)[1])[0]
output_path = os.path.join(output_path, csv_name + '.csv')
excel = ami_excel(excel_path)
print(excel_path)
print(output_path)
excel.pres_sheet.convert_amiExcelToCSV(output_path)
def main():
parser = _make_parser()
args = parser.parse_args()
_configure_logging(args)
if args.excel:
excel = ami_excel(args.excel)
if excel and excel.validate_workbook():
LOGGER.info("{}: valid".format(args.excel))
else:
if args.output:
wb = load_workbook(args.excel, data_only=True)
wb.save(args.output)
new_excel = ami_excel(args.output)
if new_excel.validate_workbook():
LOGGER.info("{}: valid".format(args.output))
else:
LOGGER.error("{}: invalid".format(args.output))
def main():
parser = _make_parser()
args = parser.parse_args()
_configure_logging(args)
if args.excel:
excel = ami_excel(args.excel)
if excel and excel.validate_workbook():
LOGGER.info("{}: valid".format(args.excel))
else:
if args.output:
wb = load_workbook(args.excel, data_only = True)
wb.save(args.output)
new_excel = ami_excel(args.output)
if new_excel.validate_workbook():
LOGGER.info("{}: valid".format(args.output))
else:
LOGGER.error("{}: invalid".format(args.output))
def check_metadata_excel(self):
if not self.metadata_files:
raise ami_bagError("Excel bag does not contain any files with xlsx extension")
bad_excel = []
for filename in self.metadata_files:
excel = ami_excel(os.path.join(self.path, filename))
if not excel.validate_workbook():
bad_excel.append(filename)
if bad_excel:
raise ami_bagError("Excel files contain formatting errors")
return True
def check_metadata_excel(self):
if not self.metadata_files:
self.raise_bagerror(
"Excel bag does not contain any files with xlsx extension")
bad_excel = []
for filename in self.metadata_files:
excel = ami_excel(os.path.join(self.path, filename))
if not excel.validate_workbook():
bad_excel.append(filename)
if bad_excel:
self.raise_bagerror("Excel files contain formatting errors")
return True
def add_json_from_excel(self):
self.excel_metadata = [filename for filename in self.data_files if os.path.splitext(filename)[1] == ".xlsx"]
for filename in self.excel_metadata:
excel = ami_excel(os.path.join(self.path, filename))
if excel.edit_sheet:
em_path = os.path.join(self.path, "data/EditMasters")
# TODO where do i error when files don't match
try:
excel.edit_sheet.add_PMDataToEM(excel.pres_sheet.sheet_values)
except:
LOGGER.error("EM's and PM's do not have 1-1 correspondence")
else:
em_filepaths = [x + ".json" for x in self.media_filepaths if em_path in x]
excel.edit_sheet.convert_amiExcelToJSON(em_path, filepaths = em_filepaths)
pm_path = os.path.join(self.path, "data/PreservationMasters")
pm_filepaths = [x + ".json" for x in self.media_filepaths if pm_path in x]
excel.pres_sheet.convert_amiExcelToJSON(pm_path, filepaths = pm_filepaths)
def set_metadata_excel(self):
self.metadata_files = [filename for filename in self.data_files if os.path.splitext(filename)[1] == ".xlsx"]
self.media_files_md = []
for filename in self.metadata_files:
excel = ami_excel(os.path.join(self.path, filename))
# collect list of filenames in metadata
if excel.pres_sheet:
paths = excel.pres_sheet.sheet_values["asset.referenceFilename"].tolist()
self.media_files_md.extend(paths)
if excel.edit_sheet:
if "asset.referenceFilename" in excel.edit_sheet.sheet_values.columns:
paths = excel.edit_sheet.sheet_values["asset.referenceFilename"].tolist()
self.media_files_md.extend(paths)
self.media_files_md = set(self.media_files_md)
return