def get_all_test_documents(self):
all_test_documents = {}
sub_directories = dataset_handler.get_all_subdirectory_names(self.data_path)
for sub_directory in sub_directories:
sub_directory_path = os.path.join(self.data_path, sub_directory)
files_indices = dataset_handler.get_names_of_files_in_directory(sub_directory_path)
for file_id in files_indices:
file_path = os.path.join(sub_directory_path, file_id)
all_test_documents[file_id] = dataset_handler.get_document_as_string(file_path,self.encoding)
return all_test_documents
def __get_all_category_directory_names(self):
subdirectories = dataset_handler.get_all_subdirectory_names(self.data_path)
subdirectories = [sd for sd in subdirectories if not sd == self.NO_CATEGORY_FOLDER_NAME]
return subdirectories