def test_download_and_extract_muscima_pp_dataset_expect_images_to_be_downloaded(
self):
destination_directory = "MuscimaPlusPlus"
downloader = MuscimaPlusPlusDatasetDownloader()
zip_file = downloader.get_dataset_filename()
number_of_images = 140
target_file_extension = "*.png"
self.download_dataset_and_verify_correct_extraction(
destination_directory, number_of_images, target_file_extension,
zip_file, downloader)
def test_download_and_extract_muscima_pp_dataset_expect_folder_to_be_created(
self):
# Arrange
destination_directory = "MuscimaPlusPlus"
downloader = MuscimaPlusPlusDatasetDownloader(destination_directory)
zip_file = downloader.get_dataset_filename()
number_of_samples_in_the_dataset = 141
target_file_extension = "*.xml"
self.download_dataset_and_verify_correct_extraction(
destination_directory, number_of_samples_in_the_dataset,
target_file_extension, zip_file, downloader)
def test_download_and_extract_muscima_pp_dataset_expect_folder_to_be_created_v2(
self):
destination_directory = "MuscimaPlusPlus"
downloader = MuscimaPlusPlusDatasetDownloader(dataset_version="2.0")
zip_file = downloader.get_dataset_filename()
number_of_samples_with_staff_lines = 140
extra_files = 1
number_of_xml_files_in_the_dataset = number_of_samples_with_staff_lines + extra_files
target_file_extension = "*.xml"
self.download_dataset_and_verify_correct_extraction(
destination_directory, number_of_xml_files_in_the_dataset,
target_file_extension, zip_file, downloader)
def test_download_extract_and_render_all_symbols(self):
# Arrange
datasetDownloader = MuscimaPlusPlusDatasetDownloader()
# Act
datasetDownloader.download_and_extract_dataset("temp/muscima_pp_raw")
image_generator = MuscimaPlusPlusSymbolImageGenerator()
image_generator.extract_and_render_all_symbol_masks("temp/muscima_pp_raw", "temp/muscima_img")
all_image_files = [y for x in os.walk("temp/muscima_img") for y in glob(os.path.join(x[0], '*.png'))]
expected_number_of_symbols = 91254
actual_number_of_symbols = len(all_image_files)
# Assert
self.assertEqual(expected_number_of_symbols, actual_number_of_symbols)
# Cleanup
os.remove(datasetDownloader.get_dataset_filename())
shutil.rmtree("temp")
help='The root directory that will contain the MuNG XMLs.')
parser.add_argument(
'-i',
'--image_root',
action='store',
default="../data/images",
help='The root directory that will contain the images of'
' scores that are represented by the MuNGs. The'
' image names must correspond to the MuNG file'
' names, up to the file type suffix.')
args = parser.parse_args()
mung_root_directory = args.mung_root
image_root_directory = args.image_root
os.makedirs(mung_root_directory, exist_ok=True)
os.makedirs(image_root_directory, exist_ok=True)
temporary_directory = "temp"
dataset_downloader = MuscimaPlusPlusDatasetDownloader()
dataset_downloader.download_and_extract_dataset(temporary_directory)
copy_tree(
os.path.join(temporary_directory, "v1.0", "data",
"cropobjects_manual"), mung_root_directory)
copy_tree(os.path.join(temporary_directory, "v1.0", "data", "images"),
image_root_directory)
shutil.rmtree(temporary_directory)
os.remove(dataset_downloader.get_dataset_filename())
os.remove(dataset_downloader.get_imageset_filename())